OBJECTIVE - Rapid-acting insulin analogs in basal-bolus regimens can reduce nocturnal hypoglycemia, so it is conceivable that twice-daily biphasic insulin analogs might reduce hypoglycemia in patients with insulin-treated type 2 diabetes. We used a continuous glucose monitoring system (CGMS) and self-reported episodes to investigate differences in the frequency of low glucose values in patients with type 2 diabetes, using either biphasic insulin aspart 30 (BIAsp 30) or biphasic human insulin 30 (BHI 30). RESEARCH DESIGN AND METHODS - This was a double-blind, two-period, cross-over trial involving 160 subjects. After 8 weeks' run-in, subjects were randomized to the first of two 16-week treatment periods. RESULTS - No differences in overall incidence of low interstitial glucose (IG) were found. Twenty-four-hour plots of CGMS showed low IG was more frequent at night than during the day and was unrecognized by patients. At night, subjects spent significantly less time (percentage of total CGMS recorded) with IG <3.5 and <2.5 mmol/l during BIAsp 30 than during BHI 30 treatment, respectively (<3.5 mmol/l: 6.36 vs. 7.93% [mean], 0.67 vs. 2.43% [median], P = 0.018; <2.5 mmol/l: 2.35 vs. 2.86% [mean], 0 vs. 0% [median], P = 0.0467). No treatment difference in A1C was observed. CONCLUSIONS - Overall rates of low glucose over 24 h were not different but were twice as frequent at night than during the day in individuals with type 2 diabetes. Compared with BHI 30, BIAsp 30 was associated with similar low IG readings over 24 h but with fewer nocturnal episodes and less self-reported nocturnal hypoglycemia.