Aims To compare blood glucose control when insulin glargine is given at lunch-time, dinner-time, and bed-time in people with Type 1 diabetes using insulin lispro at meal-times. Methods In this 16-week, three-way, cross-over study, 23 people with Type 1 diabetes were randomized to insulin glargine injection at lunch-time (L) [mean 12.37 +/- 00.34 (+/- SD) h], dinner-time (D) (18.12 +/- 00.40 h), or bed-time (B) (22.29 +/- 00.40 h), each plus meal-time insulin lispro. Each 4-week treatment period concluded with a 24-h inpatient metabolic profile. Results Insulin doses, HbA(1c), and fructosamine concentration did not differ between treatment periods. Pre-breakfast self-monitored blood glucose (SMBG) concentration was higher with injection of glargine at lunch-time than at other times [L: 9.2 +/- 0.3 (+/- SE) vs. D: 8.2 +/- 0.3 or B: 8.0 +/- 0.3 mmol/l, P = 0.016], as probably was pre-lunch SMBG (L: 8.6 +/- 0.7 vs. D: 6.4 +/- 0.7 or B: 6.4 +/- 0.8 mmol/l, P = 0.051). Pre-dinner SMBG level was higher with dinner-time glargine than other injection times (D: 9.4 +/- 0.9 vs. L: 4.9 +/- 0.9 or B: 7.4 +/- 1.1 mmol/l, P = 0.007). For 22.00 to 02.00 h, mean inpatient plasma glucose concentration was higher with injection of glargine at bed-time than other times (B: 9.1 +/- 0.6 vs. L: 7.8 +/- 0.6 or D: 6.7 +/- 0.6 mmol/l, P = 0.023). Plasma free insulin concentration was lower at the end of the afternoon with dinner-time glargine than other injection times (D: 11.5 +/- 1.4 vs. L: 20.2 +/- 1.3 or B: 16.5 +/- 1.3 mU/l, P < 0.001). Frequency of hypoglycaemia was not different, but timing of hypoglycaemia differed between treatment periods. Conclusions Blood glucose levels rise around the time of injection of insulin glargine whether given at lunch-time, dinner-time or bed-time. Bed-time injection leads to hyperglycaemia in the early part of the night which is improved by giving insulin glargine at lunch-time or dinner-time.