A Fasting Blood Sugar of 93 mg/dL Is Normal, and What the Number Does Not Tell You

A fasting blood sugar of 93 mg/dL (5.2 mmol/L) is normal. The conventional laboratory cutoffs place normal fasting plasma glucose below 100 mg/dL, impaired fasting glucose at 100 to 125 mg/dL, and diabetes at 126 mg/dL or above on two occasions. So are 95, 97, 98, and 99: all of them are inside the normal band, and none of them is “borderline” in any diagnostic sense. The differences between 93 and 98 on a single draw are mostly noise, and we will explain how much noise, because the size of the measurement error is the single most useful thing to know before you start interpreting your own numbers.

The more interesting point is that a fasting number is one sample from a continuous process, taken at the moment of the day when that process is least informative. If you want to know something about your glucose regulation, the fasting value is a weak instrument. What you want is the shape of the curve: how high you go after specific meals, how fast you come back, what your overnight baseline looks like, and how stable all of that is week to week.

How much of the difference between 93 and 98 is real

Plasma glucose from a clinical lab is measured enzymatically, usually by hexokinase or glucose oxidase on a chemistry analyzer, and the analytical coefficient of variation is roughly 2 to 3 percent. At 95 mg/dL that is about 2 to 3 mg/dL of pure instrument variation. Biological within-person variation in fasting glucose is larger, on the order of 5 to 8 percent day to day. It is driven by how well you slept, whether you fasted (coffee with milk counts), and the previous evening’s meal. Acute illness and exercise the day before also move it, as does the phase of the menstrual cycle in premenopausal women.

Combine those and the total variation on a single fasting draw is comfortably ±8 mg/dL. A person whose “true” fasting glucose is 95 will return values from roughly 85 to 105 across repeated mornings without anything changing. This is why a jump from 93 to 99 on consecutive annual physicals is not a trend, and why crossing 100 once does not mean anything on its own. Two things make a fasting value interpretable: repetition, and a pre-analytical protocol you control.

Pre-analytical error is the part people underestimate. Glucose in a tube of whole blood is consumed by red and white cells at roughly 5 to 7 mg/dL per hour at room temperature. A sample drawn at a phlebotomy center, left on a bench, and centrifuged ninety minutes later reads lower than the same blood processed immediately. Tubes with sodium fluoride (the gray-top) inhibit glycolysis, but imperfectly and slowly. If your fasting glucose came back at 93 from a lab with a long transport chain, your true value may have been 98 to 100. If you care about the number, ask for a gray-top fluoride/oxalate tube or fast processing, and hold the collection conditions constant across draws.

Why a continuous glucose monitor tells you more than any fasting draw

Continuous glucose monitors measure glucose in interstitial fluid using a subcutaneous electrochemical sensor, typically glucose oxidase coupled to a wired-enzyme electrode, sampled every one to five minutes and smoothed into a trace. The approach dates to implanted needle-type sensors paired with a wearable control unit, first demonstrated in humans in the early 1990s ¹. The modern generation is factory-calibrated, worn for 10 to 15 days, and reports on the order of 1,300 readings per day, versus one reading per year from your physical.

Two properties of the technology matter when you read your own trace. First, interstitial glucose lags plasma glucose by roughly 5 to 15 minutes, because glucose has to diffuse out of the capillaries into the interstitium. Your true post-meal peak therefore occurs a little earlier and slightly higher than the sensor shows. Second, accuracy is not uniform across the range. Sensor error, expressed as mean absolute relative difference against a reference, is typically in the 9 to 13 percent range for current devices. It worsens at low glucose and during rapid change, and calibration and error-assessment methodology has always been the hard part of the field ². Reviews of the technology and its clinical use spell out these limits and the compression artifacts that come from sleeping on the sensor ³.

The practical consequences for a healthy reader: do not treat a single CGM reading of 61 mg/dL at 4 a.m. as hypoglycemia, and do not treat a spike to 180 after a bagel as a diagnosis. Treat the trace as a distribution. Useful summary statistics over a 14-day wear are the mean, the standard deviation and coefficient of variation. Also useful are the interquartile range of the overnight period between 2 and 5 a.m., and per-meal incremental area under the curve for the two hours after each logged eating event.

A protocol we would use

If you have a fasting value of 93 and want to know something real about your glucose handling, here is the sequence we would run, from cheapest to most informative.

Start with three fasting draws on separate mornings within two weeks, same lab, same collection protocol, along with HbA1c and fasting insulin. HbA1c reflects the average glycation of hemoglobin over the preceding roughly 90 days, weighted toward the most recent month. Fasting insulin is the one number that tends to move before fasting glucose does: a normal glucose maintained by a rising insulin level is a different physiological state from a normal glucose maintained by a low one, and both read as “93.”

Then wear a CGM for 14 days while eating your normal diet and logging meals with timestamps. Do not change what you eat during the first wear. The purpose of wear one is to establish a baseline distribution. It is not the time to test hypotheses. Export the raw data rather than reading only the vendor app summary. Most vendors provide a CSV with a timestamp column and a glucose column at five-minute intervals, and you can compute the statistics yourself:

import pandas as pd
d = pd.read_csv("cgm.csv", parse_dates=["timestamp"]).set_index("timestamp")
g = d["glucose_mgdl"].resample("5min").mean().interpolate(limit=3)
print(g.mean(), g.std(), 100*g.std()/g.mean())
night = g.between_time("02:00", "05:00")
print(night.quantile([0.25, 0.5, 0.75]))

On a second wear, run standardized challenges. The same test meal (a fixed 50 g or 75 g carbohydrate load, weighed, same time of morning, after a 10-hour fast) repeated three or four times gives you an estimate of your own within-person variability for that meal, which is typically large enough to swamp the difference between two different foods tested once each. Only after you have that variance estimate can you compare foods meaningfully.

One caution about symptoms. People without diabetes sometimes feel shaky, sweaty, or foggy at glucose values in the 70s and 80s that are physiologically normal, and the sensation tracks the rate of fall and the adrenergic response more than the absolute number. The perception of hypoglycemia is unreliable in both directions, and the experience has been studied as something people learn to read from a mix of bodily signals and device numbers rather than from the value alone ⁴. Recurrent shakiness, confusion, fainting, or documented values below 55 mg/dL on a laboratory glucose need a clinician, because the workup for true hypoglycemia in a non-diabetic person is specific and not something to attempt from CGM data. The same applies to any fasting value at or above 126, or an HbA1c at or above 6.5 percent.

Questions people also ask

What is my A1C if my blood sugar is 93? There is no reliable conversion from a single fasting value. The common regression-based estimate, mean glucose ≈ 28.7 × A1C − 46.7, uses 24-hour mean glucose, not fasting glucose, and a mean of 93 would correspond to roughly 4.9 percent. Fasting glucose typically sits below your 24-hour mean if you eat carbohydrate, so a fasting 93 is compatible with an A1C anywhere from about 4.8 to 5.6 percent. Measure the A1C.

Is 97 borderline diabetic? No. Impaired fasting glucose starts at 100 mg/dL and diabetes at 126 mg/dL, and 97 is within measurement noise of 93. That said, epidemiological work has found that risk of later type 2 diabetes rises progressively across the 90 to 99 mg/dL band compared with lower values, which is a statement about population risk gradients and not about your individual diagnosis ⁵.

What happens when your blood sugar is 90? Nothing. It is a normal value at any time of day, and in a person without diabetes the counter-regulatory response to falling glucose does not begin until roughly the high 60s.

Is a 93 sugar level normal for diabetics? It is within the range many treatment plans target before meals, but interpretation in someone on glucose-lowering medication depends entirely on the regimen and belongs with the prescribing clinician. Real-time CGM is used partly because seeing the trend line reduces the fear of going low, an effect documented in qualitative work with older adults on insulin ⁵.

Why do I feel shaky but don’t have diabetes? Shakiness with normal glucose is common and has many explanations, including a rapid fall from a post-meal peak, caffeine, poor sleep, anxiety, and simple adrenergic sensitivity. A CGM worn while you log the symptom episodes will tell you whether the feeling coincides with a low value, a fast drop, or neither. If episodes are frequent or severe, see a physician rather than trying to interpret sensor data on your own, since sensor accuracy degrades exactly in the low range where you would want to trust it ⁶.

Footnotes

1. V. Poitout, D. Moatti-Sirat, G. Reach, et al. A glucose monitoring system for on line estimation in man of blood glucose concentration using a miniaturized glucose sensor implanted in the subcutaneous tissue and a wearable control unit. Diabetologia, 1993. https://doi.org/10.1007/bf00404077 ↩

2. Volker Lodwig, Lutz Heinemann. Continuous Glucose Monitoring with Glucose Sensors: Calibration and Assessment Criteria. Diabetes Technology & Therapeutics, 2003. https://doi.org/10.1089/152091503322250596 ↩

3. David C. Klonoff, David Ahn, Andjela Drincic. Continuous glucose monitoring: A review of the technology and clinical use. Diabetes Research and Clinical Practice, 2017. https://doi.org/10.1016/j.diabres.2017.08.005 ↩

4. Annemarie Mol, John Law. Embodied Action, Enacted Bodies: the Example of Hypoglycaemia. Body & Society, 2004. https://doi.org/10.1177/1357034x04042932 ↩

5. Michelle L. Litchman, Nancy A. Allen. Real-Time Continuous Glucose Monitoring Facilitates Feelings of Safety in Older Adults With Type 1 Diabetes: A Qualitative Study. Journal of Diabetes Science and Technology, 2017. https://doi.org/10.1177/1932296817702657 ↩ ↩²

6. P. Avari, M. Reddy, N. Oliver. Is it possible to constantly and accurately monitor blood sugar levels, in people with Type 1 diabetes, with a discrete device (non‐invasive or invasive)?. Diabetic Medicine, 2019. https://doi.org/10.1111/dme.13942 ↩