What is the distribution per hour of Humalog action?

I used to know this, but have forgotten how much (presuming my duration is 5 hours) action I get during hours 1, and 2. PLEASE don’t say it is 1/5 of the total, per hour, as that is so wrong. I can’t find the spreadsheet where I had this all worked out long ago.

Never mind! I found the activity curve.

Can you share what you’ve found?

For me, the reference is page 18

1 Like

Back before pumps calculated insulin on board (IOB), I learned how to use manually calculated insulin corrections that considered IOB. Since this was all “back of an envelope” street math, I used the crude linear analysis of 1/5 of the total insulin being used per hour. I was aware of the imprecision but its simplicity helped me during this time.

This is something John Walsh recommended in one of his earlier book editions back in the '90s. It worked well enough for me but I’m more than happy to now use the more precise calculations produced by calculus equations that live in my diabetes devices.

there is so much more insulin action around the 1.5 to 2.5 hour. Perhaps I should explain a bit more with regards to my need to remember the distribution. In the morning I might be spot on, or I’m as much as 30 points higher than I should be. So I’ll bolus, then sit back and wait until I get down to around 115-125 and then eat breakfast (I’m not on a schedule, so I can eat "whenever). If I’m not paying attention to my bgs, I’ll get too low before I eat (but not hypo–just “too low” for the amount of IOB) and then I have to eat a second time. Since my AM bolus varies by up to a couple of units, I need to be a bit more “scientific” about the expected time it takes me to drop to 110-115, so that I can set a timer on my phone instead of glancing over and over again at the CGM. I get busy doing other things and don’t always look at the point that I’ve hit the range I want. Make sense?

I’ve got Walsh’s newest book–I really dig the spiral binding and the inclusion of CGM chapters that weren’t available when I got my edition around 1996.

Thanks Dave! I remember seeing that before. I find the most effective time for me is between an hour and a hour and a half and then I really start to taper off. For me I find prebolusing half the dose a half hour before I eat works well.

I love to prebolus and get irritated at myself when I forget as once I go up, it is harder to come back down! And not forgetting to eat is the other trick because then I am drinking OJ instead of eating! I love Oj so not so bad!

But looking at the PDF made me hungry and I skip eating in the morning because my sugars are too hard to control in the AM. But man there was a picture of a milkshake that got me the most! Luckily there is no ice cream in the house! (vegan lol, there might be ice cream but I don’t eat it)

Same for me, Marie. I usually pre-bolus.

1 Like

Isn’t the curve on page 11 of the PDF the answer you’re looking for?

that page and page 18 don’t look much different for humalog. but the description is a bit different for the duration. I think they made a bit of a mistake if u compare the 2 purple curves. the first one shows a duration of ~4 and the one on page 18 says 4-5 hours. The graphs look pretty darn similar to me, but you’d think the graph on page 18 would be longer.

If someone has a better graph, I’d like to see it.

Here is a reference with time ranges.

My favorite, most unpredictable insulin:
NPH (Novolin N, Humulin N, ReliOn)
Onset 2-4 hours
Peak 4-8 hours
Duration 10-18 hours


I think it’s the same graph. Looking at the baseline, I’d say all but a small amount of Humalog activity is gone by hour 4. I know that’s what my endo told me.

I agree that using a simple square wave (1/5 per hour for 5 hours) is all but worthless. The curve shows thats just not true. You can probably ball-park estimate the activity by hour using the curve.

At least for me, my physical activity level has a huge impact on the insulin timeline. When I’m sedentary it takes what seems like hours before I see a significant response to a bolus. When I’m very active I can see a significant response in just a few minutes. Just another variable to deal with.


This Journal of Diabetes Science and Technology paper addresses the issue of insulin’s action over time. Rapid acting analog insulin’s duration is commonly thought to be in the 5-6 hour range.

Here’s a graph from that paper that shows the percent remaining insulin on board (IOB) with different duration assumptions.

Using this chart, if a five-hour duration of insulin action is assumed, here’s the approximate %IOB over five hours and the distribution per hour.

1 hour - 90% IOB or 10% used this hour
2 hours - 70% IOB or 20% used this hour
3 hours - 35% IOB or 35% used this hour
4 hours - 12% IOB or 23% used this hour
5 hours - 0% IOB or 12% used this hour


For me, it lasts 5 to 5.5 hours. The endo doesn’t know how long it will last in a particular person.

thanks Terry, but that doesn’t address my issue which is for me to know how much of a drop I’ll get in the first and second hours. IOB by itself, doesn’t equal the blood-lowering amounts we get. If it were that simple, I’d just look at my pump’s IOB. That’s why I was look for, and posted the graph.

Sorry – I took the title of your post literally. So, you’re not asking what percentage of a total insulin dose is active each hour. Determining the actual blood glucose drop per hour is tricky.

Factors would include your relative insulin sensitivity as driven by many factors including recent exercise, and sleep quality. Other factors include the quality and mix of food eaten, insulin dose size and timing. There’s no standard chart that could show that. That’s where the art of insulin dosing comes into play.

Right. I want to know how much of a drop occurs for hours 1 and 2, per unit (my ISF is 30,so I figure it is somewhere around ~6 for the first hour and about ~14 for the second. Agreed it is tricky, but I used to know these numbers like the back of my hand but over time have forgotten to utilize them to avoid bg excursions during bolusing.

I was concerned here with the drop PRIOR to eating, as I’m bolusing well ahead of breakfast.

I can’t get that pdf to display.

If you’re interested in the BG drop during your pre-bolus time, then that would be a simple calculation. If you took a 10 unit meal dose (round number for illustrative purposes), during the first hour, 10% of the total insulin dose (1 unit) would drop your pre-dosing BG by 30 mg/dl (1.7) using your ISF of 30. During the second hour your BG would continue to drop another 60 mg/dL or 3.3 mmol/L (20% of 10 units = 2 units and ISF of 30).

I’m using the chart numbers I cited above and this is an ideal analysis and doesn’t account for any other outside influences like exercise or dawn phenomema. A basal rate set too high or low would also distort this analysis.