Normal Blood Sugar After Eating, and How to Measure Your Own

Oak
glucose cgm postprandial metabolic-health data-analysis
Huge long-necked grazing animals in a misty valley, each with a glowing stripe along its flank at a different brightness.

In a person without diabetes, glucose before a meal usually sits between 70 and 99 mg/dL (3.9–5.5 mmol/L), peaks 45 to 75 minutes after the first bite, and is back near baseline by two hours. The conventional non-diabetic ceiling at two hours is under 140 mg/dL (7.8 mmol/L). The peak itself, which a fingerstick at two hours almost always misses, commonly lands between 110 and 160 mg/dL depending on the meal, and brief excursions above 140 are ordinary in healthy people. For someone with diagnosed diabetes, the usual clinical targets are 80–130 mg/dL before meals and under 180 mg/dL two hours after the start of a meal, but targets are individual and set with a clinician, not from a web page.

Two conversions, because search queries mix units: 8 mmol/L after eating is 144 mg/dL. 200 mg/dL is 11.1 mmol/L. An “87 glucose level after eating” is a normal fasting-range value that either means you tested late, ate very little carbohydrate, or caught the post-peak undershoot.

What continuous monitoring shows in people without diabetes

Fingerstick guidance was built around a sparse sampling schedule: fasting, and two hours post-meal. CGM samples every five minutes and shows a different picture. In healthy cohorts wearing factory-calibrated sensors under free-living conditions, mean glucose clusters near 100 mg/dL and the large majority of the day is spent between 70 and 140 mg/dL, with short, meal-linked departures above that band 1. Early CGM work in healthy volunteers eating ordinary and standardized meals found the same shape: tight baseline, sharp rise, return within roughly two hours, and maximum values that depend heavily on what was eaten 2.

More recent CGM data in people without diabetes shows that post-meal peaks above 140 mg/dL occur in a meaningful fraction of normal meals, especially high-glycemic ones, without any implication of disease 3. The useful signal is not whether you ever cross 140. It is how high you go, how long you stay there, and how reproducible the response is for a given meal.

Three descriptors carry almost all the information in a meal response:

  • Peak (Cmax) and time to peak. Healthy responses peak early, typically 45–75 minutes. A late peak (>90 minutes) with a slow decline is the more interesting pattern.
  • Two-hour delta: glucose at t+120 minus pre-meal baseline. Near zero is the healthy default.
  • Incremental area under the curve over three hours (iAUC), which captures both height and duration in one number.

Where 140 and 180 came from

The 140 mg/dL threshold is the two-hour value of the 75 g oral glucose tolerance test: below 140 is normal, 140–199 is impaired glucose tolerance, 200 or above on two occasions is a diabetes diagnosis. That number describes a specific standardized challenge in a fasted state, measured on venous plasma. It is not a rule that your glucose must never exceed 140 after a burrito.

The 180 mg/dL target for people with diabetes is a practical ceiling from intensive-therapy trials, chosen as a level achievable without unacceptable hypoglycemia 4. It is also the upper bound of the standard CGM “time in range” band, 70–180 mg/dL, which drives most modern diabetes device metrics 5. Neither threshold is a biological cliff. They are decision boundaries, and which one applies to you is a clinical question.

Your sensor’s error budget

Before drawing conclusions from a curve, know what the curve is.

CGM measures glucose in interstitial fluid, not blood. There is a physiological lag plus a filtering delay, together roughly 5 to 15 minutes, largest when glucose is changing fastest, which is exactly during a meal spike 6. Your recorded peak is therefore later and usually slightly lower than the true plasma peak.

Sensor accuracy is expressed as MARD, mean absolute relative difference against a reference. Current consumer sensors report MARD around 9%. At 140 mg/dL that is roughly ±13 mg/dL of typical error, with tails well beyond it. Day-one and final-day readings are often the worst, and compression during sleep produces false lows that look like nocturnal hypoglycemia but resolve the moment you roll over 7. Treat any single reading as an estimate with a wide interval. Treat a repeated pattern across three meals as real.

Also know what does not work: optical and PPG-based “non-invasive glucose” from a watch or phone camera. The measurement problem is unsolved at clinical accuracy, and published smartphone-PPG models are research demonstrations, not devices you should build a decision on 89.

A meal-response protocol worth running

Free-living CGM data is confounded by everything. If you want comparable numbers, standardize.

  1. Start each test fasted 10–12 hours, or at least 4 hours after the previous meal, same time of day. Morning insulin sensitivity differs from evening.
  2. Fix the meal exactly: same grams, same preparation, same eating order, consumed in under 15 minutes. Log the first-bite timestamp to the minute.
  3. No exercise 2 hours before or 3 hours after. A 15-minute walk after eating measurably flattens the curve, which is informative but destroys comparability if it is uncontrolled 1.
  4. Repeat each meal three times on nonconsecutive days. Within-person variability for the same meal is large enough that n=1 tells you little.
  5. Keep a fingerstick meter and test strips on hand to sanity-check any reading that surprises you.

A reasonable panel: 75 g glucose equivalent as a reference challenge, one high-refined-carbohydrate meal, one mixed meal with protein and fat, and one fat-and-protein-dominant meal. The reference challenge gives you an anchor comparable to published data.

Getting the numbers out of the export

Dexcom Clarity exports CSV with Timestamp (YYYY-MM-DDThh:mm:ss), Event Type, and Glucose Value (mg/dL). Values below range appear as the string Low, above range as High. Abbott’s LibreView CSV uses Device Timestamp in local format plus separate columns for historic (15-minute) and scan glucose. Parse both to a single tidy frame of (timestamp, mg_dl) before doing anything else.

import pandas as pd, numpy as np

df = pd.read_csv("clarity.csv", skiprows=range(1, 11))
df = df[df["Event Type"] == "EGV"].copy()
df["t"] = pd.to_datetime(df["Timestamp (YYYY-MM-DDThh:mm:ss)"])
df["g"] = pd.to_numeric(
    df["Glucose Value (mg/dL)"].replace({"Low": 39, "High": 401}), errors="coerce"
)
df = df.dropna(subset=["g"]).set_index("t").sort_index()

def meal_metrics(df, first_bite, window_min=180):
    t0 = pd.Timestamp(first_bite)
    w = df.loc[t0 - pd.Timedelta("20min") : t0 + pd.Timedelta(minutes=window_min), "g"]
    base = w.loc[t0 - pd.Timedelta("20min") : t0].mean()
    post = w.loc[t0:]
    mins = (post.index - t0).total_seconds() / 60
    inc = np.clip(post.values - base, 0, None)
    return {
        "baseline": round(base, 1),
        "peak": round(post.max(), 1),
        "t_peak_min": int(mins[post.values.argmax()]),
        "delta_2h": round(post.asof(t0 + pd.Timedelta("2h")) - base, 1),
        "iauc_180": round(np.trapz(inc, mins), 1),   # mg/dL * min
        "min_3h": round(post.min(), 1),
    }

Report iAUC in mg/dL·min and always alongside baseline, since incremental area hides a high starting point. Add min_3h because a deep post-peak undershoot below baseline is one of the more reproducible patterns worth showing a clinician.

For whole-wear summaries, compute coefficient of variation (SD/mean), percent of time in 70–140 and 70–180, and mean glucose. CV is the stability metric that survives sensor bias better than absolute values do.

What one high reading means

A single 200 mg/dL two hours after a large high-carbohydrate meal, on a sensor with 9% MARD, is one data point. A reading of 350 is outside anything a meal alone typically explains in a person without diabetes, and belongs in front of a clinician the same day, not in a notebook. Persistent post-meal values above 140 across repeated standardized meals, or fasting values repeatedly at or above 100, are a reason to get laboratory testing: venous fasting glucose, HbA1c, and possibly an OGTT. CGM is a measurement instrument, not a diagnostic one, and no threshold here substitutes for that workup.

Questions people also ask

Is 140 blood sugar high after eating? At two hours, 140 mg/dL is the upper edge of the normal reference and worth repeating. As a transient peak at 60 minutes, it is common in people without diabetes, particularly after refined carbohydrate 3.

Is 200 blood sugar normal right after eating? No. In people without diabetes, 200 mg/dL is above typical peaks even for high-glycemic meals 2. One sensor reading can be error. A repeated pattern calls for laboratory confirmation with a clinician.

Why do people with diabetes wake up around 3 a.m.? Overnight CGM traces usually show either a nocturnal low or a pre-dawn rise driven by cortisol and growth hormone. Distinguishing them requires the trace, and compression artifacts from lying on the sensor mimic lows convincingly 7.

What is the normal sugar level by age after food? Reference ranges do not change materially with age in healthy adults. Clinical targets are sometimes loosened in older adults with diabetes because hypoglycemia risk outweighs the benefit of tight control, which is a clinician’s judgment call 10.

What is the ideal blood sugar after dinner? Use your own baseline as the reference: back within about 20 mg/dL of pre-meal by two hours, with an early peak. Evening responses are often higher than morning responses to the identical meal, so compare dinner to dinner.

Oak builds longitudinal molecular profiles of individuals: whole-genome sequencing, RNA sequencing, proteomics, blood biomarkers, and continuous glucose data, integrated into one model of you. Build your profile.

Footnotes

  1. Stephanie N. DuBose, Zoey Li, Jennifer L. Sherr, et al. Effect of Exercise and Meals on Continuous Glucose Monitor Data in Healthy Individuals Without Diabetes. Journal of Diabetes Science and Technology, 2020. https://doi.org/10.1177/1932296820905904 ↩ ↩2

  2. Guido Freckmann, Sven Hagenlocher, Annette Baumstark, et al. Continuous Glucose Profiles in Healthy Subjects under Everyday Life Conditions and after Different Meals. Journal of Diabetes Science and Technology, 2007. https://doi.org/10.1177/193229680700100513 ↩ ↩2

  3. Paul R.E. Jarvis, Jessica L. Cardin, Pamela M. Nisevich-Bede, et al. Continuous glucose monitoring in a healthy population: understanding the post-prandial glycemic response in individuals without diabetes mellitus. Metabolism, 2023. https://doi.org/10.1016/j.metabol.2023.155640 ↩ ↩2

  4. Bernard Zinman. Glucose control in type 1 diabetes: From conventional to intensive therapy. Clinical Cornerstone, 1998. https://doi.org/10.1016/s1098-3597(98)90016-3 ↩

  5. Bruce W. Bode. Clinical Utility of the Continuous Glucose Monitoring System. Diabetes Technology & Therapeutics, 2000. https://doi.org/10.1089/15209150050214104 ↩

  6. R. Hovorka. Continuous glucose monitoring and closed‐loop systems. Diabetic Medicine, 2005. https://doi.org/10.1111/j.1464-5491.2005.01672.x ↩

  7. Lisa K. Gilliam, Irl B. Hirsch. Practical Aspects of Real-Time Continuous Glucose Monitoring. Diabetes Technology & Therapeutics, 2009. https://doi.org/10.1089/dia.2008.0135 ↩ ↩2

  8. Sandeep Kumar Vashist. Non-invasive glucose monitoring technology in diabetes management: A review. Analytica Chimica Acta, 2012. https://doi.org/10.1016/j.aca.2012.03.043 ↩

  9. Gaobo Zhang, Zhen Mei, Yuan Zhang, et al. A Noninvasive Blood Glucose Monitoring System Based on Smartphone PPG Signal Processing and Machine Learning. IEEE Transactions on Industrial Informatics, 2020. https://doi.org/10.1109/tii.2020.2975222 ↩

  10. Rabbi Swaby, Tabitha Randell. Strategies for optimising blood glucose control in diabetes mellitus. Paediatrics and Child Health, 2021. https://doi.org/10.1016/j.paed.2021.01.004 ↩