Background We examined how habitual sleep duration interacts with recent sleep (2 nights) to predict morning oral glucose tolerance test results. We hypothesized that short habitual and recent sleep durations would be additive for poor glucose control. Methods A biracial population of adults (n = 1559) without known diabetes and recruited from the workforce of 2 urban universities was assessed for glycated hemoglobin and underwent oral glucose tolerance testing. We used plasma 2-hour postloading (75 g) measurements. Participants answered sleep questions using 30-minute forced-choice formats. We employed multivariable logistic regression to derive odds ratios. Results Shorter habitual sleep duration was associated with greater odds ratios of glycated hemoglobin ≥6.0% increasing by 30-minute intervals beginning at < 7.0 hours and were more pronounced as durations shortened. Among participants with glycated hemoglobin < 6.0% and < 7.0 hours of habitual sleep (n = 636), abnormal glucose tolerance (2-hour oral glucose tolerance test ≥140 mg/dL) was significantly associated with a total sleep duration of ≤11 hours the 2 nights preceding oral glucose tolerance testing, but was not associated with longer sleep durations. Results were independent of age, sex, race, body mass index, smoking, history of cardiovascular disease, or use of antihypertensive or cholesterol-lowering medication. Additional analyses implied that longer-than-usual recent sleep durations were protective for abnormal oral glucose tolerance testing. Discussion Short habitual and recent sleep durations interact in predicting abnormal glucose on oral glucose tolerance testing. Self-reported data are sufficiently sensitive to reflect 30-minute differences in sleep between individuals. Future studies examining other aspects of sleep, such as perceived sleep quality and objectively measured sleep duration and architecture, would be necessary to confirm these findings. Conclusions Short sleep duration for 2 nights prior to morning oral glucose tolerance testing may elevate glucose levels, this effect being detected among individuals habitually obtaining < 7 hours sleep and obtaining ≤11 hours of sleep for 2 nights preceding testing.