Which Flies Catch the Most Trout? 514 Fish From 193 Logged Fly Fishing Trips

Eighteen streamer trips. Twenty-two fish. That's the ugliest line in eight years of my fishing log, and I didn't know it was there until I finally added the column up. I'd pulled 193 logged fly fishing trips out of Bushwhack to settle a question I was sure I already knew the answer to: which flies catch the most trout, and by how much.

What's actually in this dataset

The short answer first. Across 193 logged fly fishing trips and 514 fish, the Elk Hair Caddis produced 3.19 fish per trip over 16 trips, 20% above my overall average of 2.66. The Hippie Stomper, the single pattern I fish more than any other at 21 trips, produced 2.24, which is 16% below that same average. The fly I reached for most was quietly the worse fly.

Every number below comes from one angler's log: mine. It covers 193 logged fly fishing trips and 514 fish landed, running from 28 April 2018 to 21 September 2026. That works out to an overall average of 2.66 fish per trip, which is the baseline every fly and every style in this post gets measured against.

A few things you should know before you quote any of it.

Those 193 trips are the fly fishing subset of a larger log. I've logged 221 trips in total across all methods, so the spin and bait days are excluded here and none of the 514 fish are conventional-tackle fish. Waters are mostly Utah, with trips to the Provo River, the Green River, the Nestucca River in Oregon, and the Kenai River in Alaska. I'm deliberately not naming my small home waters, which is where the bulk of the trips happened, so treat the water-by-water numbers as directional rather than a place recommendation.

Fly attribution works per fish, not per trip. On a dry-dropper rig the log records the dropper and the dry as separate entries, and each fish is credited to whichever fly it actually ate. So the "trips" column for a fly means trips that fly was on the leader, and those counts overlap. They don't add up to 193. Two style labels differing only by capitalization were merged into a single Dry Dropper row before any of this was calculated.

Which flies catch the most trout, sorted by fish per trip

Here are the eight patterns that accounted for the most fish. Together they landed 267 of the 514 fish, just under 52% of everything in the log. I've sorted by fish per trip rather than raw total, because raw total mostly measures how often I reach for something.

Fly	Trips	Fish	Fish per trip	vs 2.66 avg	Sample
Tenkara Fly (red wet fly)	5	37	7.40	+178%	Tiny
Perdigon (pink nymph)	4	25	6.25	+135%	Tiny
Elk Hair Caddis (brown dry)	16	51	3.19	+20%	Solid
Copper John (red nymph)	7	22	3.14	+18%	Small
Rainbow Warrior (red nymph)	8	22	2.75	+3%	Small
Pheasant Tail (brown nymph)	14	37	2.64	-1%	Solid
Glo Bug (pink nymph)	10	26	2.60	-2%	Moderate
Hippie Stomper (black dry)	21	47	2.24	-16%	Best in set

Read that table from the bottom up and you'll see why I led with those two patterns. They carry the two largest samples in the set, 21 trips and 16 trips, and they sit on opposite sides of my average by a combined 36 percentage points. That's the one comparison here built on enough trips to lean on.

The two flies at the top of the table are the ones you should trust least. I'll come back to why.

The Elk Hair Caddis is the most trustworthy number in the whole log

Fifty-one fish across sixteen trips, and that row is the brown Elk Hair Caddis only. Olive and tan versions appear in my log as separate entries, including a cutthroat session on a coastal creek in July, and not one of those fish is counted in the 3.19. I'm keeping them out on purpose, because a section arguing for sample integrity shouldn't quietly pool three flies into one average.

What makes that row convincing is that no single session props it up. Inside those sixteen trips it produced browns on a small freestone creek in April with the water at 49 degrees, rainbows and tiger trout off a mountain reservoir in June, and bluegill out of an urban canal in January with the air at 39 degrees.

That spread is the point. A fly with a high average built from one water in one month is a fly that got lucky. A fly with a high average built from three water types and four species across three separate months is closer to a real edge.

It also lines up with what's in the drift. Caddisflies are one of the dominant aquatic insect groups in western streams, and Utah State University Extension notes that macroinvertebrates "provide an important food source for fish and other predators." I'm not claiming my log proves an entomological point. I'm saying the pattern that performs best in my data imitates the bug that's most reliably available, and that's a satisfying place for the numbers to land.

You might also enjoy: Kenai River Bead Fishing: The Rig, the Regs, and the Runs

Why does the dry-dropper account for 64% of the fish?

Because I fish it constantly. That's most of the answer, and it's less exciting than it looks.

Style	Trips	Fish	Fish per trip	Share of all fish
Tenkara	10	55	5.50	10.7%
Dry Dropper	115	328	2.85	63.8%
Dry Fly	22	49	2.23	9.5%
Nymphing	23	51	2.22	9.9%
Hopper Dropper	5	9	1.80	1.8%
Streamer	18	22	1.22	4.3%

Dry-dropper took 63.8% of the fish off 59.6% of the trips. Strip the volume out and its actual efficiency edge is 2.85 fish per trip against a 2.66 average, a margin of just 7%. It's not a cheat code. It's a rig I trust enough to leave on, and leaving it on is what produced the enormous raw total.

Tenkara is the number that jumps off the table at 5.50 fish per trip, more than double the baseline. Hold that thought too.

Nymphing sits below average at 2.22, which surprised me until I looked at where those trips happened. Nymphing is what I default to on unfamiliar or difficult water, including the Kenai River, where a bead under an indicator produced nine pink salmon in an afternoon in August 2026. That session was a personal best for the species. It's also the kind of trip I'd never have taken a dry-dropper on.

Streamers are the clearest underperformer in eight years of logs

Eighteen trips, 22 fish, 1.22 per trip. That's 54% below my overall average. Had those eighteen trips merely hit the baseline they'd have produced about 48 fish instead of 22, so the streamer habit has cost me roughly 26 fish over eight years.

Worse, and this is the stat I find hardest to argue with: not one streamer session appears among the 30 best-producing entries in the log. Every one of those top sessions landed four or more fish. Not a single streamer day got there.

Now, the honest counterweight. Streamer anglers aren't chasing numbers, they're chasing size, and my log's fish-per-trip column is blind to that. A four-fish caddis afternoon of eight-inch browns outranks a one-fish streamer day, even if that one fish was the best trout of the year. If you fish streamers for big browns, fish-per-trip is the wrong scoreboard and you should ignore this section entirely.

But I don't fish streamers for big browns. I fish them because I get bored, and the water looks streamer-y, and stripping feels productive in a way that watching a dry fly doesn't. Eight years of data says that feeling is lying to me. Every guide I've fished with will tell you presentation beats pattern, and my streamer numbers are almost certainly a presentation problem rather than a fly problem. Either way the fix is the same: I need to get better at it or stop doing it on days I actually want to catch fish.

Do these numbers prove one fly is better than another?

No. And the gap between "this fly produced" and "this fly causes fish" is where almost every fly-data article on the internet goes wrong, so let me be specific about the four ways my own table could be fooling me.

There's no randomization. The only decent controlled experiment I know of on this question is the CRACKPOT trial, published in the BMJ in 1998, a randomized Latin square in which five anglers fished five randomly ordered dry flies across 25 sessions of up to five hours each. They landed 72 fish. The Cinnamon Sedge took 20 of them, the Black Gnat just 7, and the authors calculated a probability of 0.025 for a result that low by chance. That's 72 fish under randomized conditions. I have 514 fish under no conditions at all: I tied on whatever I felt like, whenever I felt like it. More data does not beat a better design.

Reverse causation is everywhere. I reach for a Perdigon when the water carries a little color and I already expect a good day. The fly didn't create those conditions, it just happened to be knotted on during them. Any fly I save for promising water will look like a great fly.

There's no time denominator. This is the one that bothers me most. The log counts trips, not hours per pattern. A fly that isn't working gets cut off after twenty minutes; a fly that's working stays on for four hours. So an underperforming pattern is partly being punished for the fact that I gave up on it, and a producing pattern is partly being rewarded for the fact that I didn't. Fish per trip quietly bakes my own impatience into the result.

You might also enjoy: Fly Fishing Gifts Under $50: 11 Amazon Picks Sorted by Job

The small samples are single days wearing a disguise. Look hard at the top two rows of the fly table:

• Perdigon: 25 fish, but 21 of them came on one afternoon in March 2026 on a small Utah freestone creek, 52 degrees, mostly cloudy, slightly stained water. Remove that single session and the Perdigon's remaining record is 4 fish across 3 trips, which is 1.33 per trip. Its 6.25 average is one good day and almost nothing else.
• Tenkara Fly: 37 fish, but 32 of them came in three sessions over two days in July 2024. The rest of its history is 5 fish across 2 trips.

The Tenkara style number has the same problem. Four sessions between 25 and 27 July 2024 produced 40 of its 55 fish. Those were high-altitude brook trout in midsummer, and brook trout at altitude will eat a cigarette butt on a good morning. Take that stretch out and tenkara drops to 2.50 fish per trip, right in line with everything else. It was never a rod advantage. It was a brook trout advantage.

Which is exactly why I published the sample-size column instead of quietly leaving it out.

The famous rivers lost to the little creek

One more result worth putting on the table, with the same caveats attached. The Provo River is Utah's best-known trout fishery and it's my worst-performing named water: 19 trips, 30 fish, 1.58 per trip, roughly 41% below my average. The Green River sits at 2.33 across 6 trips. The Nestucca River in Oregon is the best of the named waters at 3.40 across 5 trips.

Meanwhile the small unnamed freestone creek I fish most produced 183 fish across 60 trips, 3.05 per trip.

I don't think this means the Provo is a bad river. It means the Provo has educated fish and heavy pressure, I fish it a handful of times a year without learning it properly, and I've spent eight years learning one small creek intimately. That's an argument for depth over novelty, and it's the single most useful thing eight years of logging has taught me. Utah's rules are water-specific and change, so check the Utah Division of Wildlife Resources before fishing anywhere new.

What I actually changed

The Hippie Stomper came out of rotation as a searching pattern. Twenty-one trips at 16% below my own average is enough evidence for me, and I'd been fishing it mostly because it's easy to see. Visibility is a real feature. It just isn't worth a 16% haircut, and the Elk Hair Caddis in brown is nearly as visible in the sizes I fish.

The Elk Hair Caddis is now the default first fly on unfamiliar water.

Streamers are now a deliberate choice for a specific reason, not a bored default. And I started logging time-on-fly, not just fly, so that next time I run these numbers the biggest confound in the dataset is gone.

None of this transfers cleanly to your water. My 2.66 baseline is a function of small Utah creeks, a lot of eight-inch browns, and my own mediocre streamer skills. The transferable part is the method: log the fly, log the fish, and divide. If you want the companion piece, I ran the same treatment on my blank days in what percentage of fishing trips are skunked, and if you'd rather build your own dataset than borrow mine, this guide to starting a fishing log covers what to record. You can also try Bushwhack against a demo account to see what eight years of this looks like before logging a single trip of your own.

Two hundred trips from now I'll run it again, and I fully expect the Elk Hair Caddis to have regressed toward the mean along with everything else.