This observational pilot study examined the association between diet, meal pattern and glucose over a 2-week period under free-living conditions in 26 adults with dysglycemia (D-GLYC) and 14 with normoglycemia (N-GLYC). We hypothesized that a prolonged eating window and late eating occasions (EOs), along with a higher dietary carbohydrate intake, would result in higher glucose levels and glucose variability (GV). General linear models were run with meal timing with time-stamped photographs in real time, and diet composition by dietary recalls, and their variability (SD), as predictors and glucose variables (mean glucose, mean amplitude of glucose excursions [MAGE], largest amplitude of glucose excursions [LAGE] and GV) as dependent variables. After adjusting for calories and nutrients, a later eating midpoint predicted a lower GV (β = −2.3, SE = 1.0, p = 0.03) in D-GLYC, while a later last EO predicted a higher GV (β = 1.5, SE = 0.6, p = 0.04) in N-GLYC. A higher carbohydrate intake predicted a higher MAGE (β = 0.9, SE = 0.4, p = 0.02) and GV (β = 0.4, SE = 0.2, p = 0.04) in N-GLYC, but not D-GLYC. In summary, our data suggest that meal patterns interact with dietary composition and should be evaluated as potential modifiable determinants of glucose in adults with and without dysglycemia. Future research should evaluate causality with controlled diets.