The relevance of gemcitabine timing for chronotherapeutic optimisation was investigated. Healthy mice received multiple doses of gemcitabine (120, 160 or 200 mg kg(-1) injection (inj)(-1)) at one of six circadian times (3, 7, 11, 15, 19 or 23 h after light onset--HALO) on days 1, 4, 7 and 10 or a single dose of gemcitabine (400 mg kg(-1)) at 11 or 23 HALO+/-cisplatin (5 mg kg(-1) at 1 min, 4 or 8 h later). Mice bearing Glasgow osteosarcoma received multiple doses of gemcitabine (200 mg kg(-1) inj(-1)) at 11 or 23 HALO+/-cisplatin (5 mg kg(-1) inj(-1) at 1 min or 4 h later) on days of 10, 13, 16 and 19 following tumour inoculation. A circadian rhythm in body weight loss was statistically validated, with 1030 HALO corresponding to the least toxic time (95% CL, 0800 to 1300). Gemcitabine dosing produced least body weight loss and least neutropenia after injection at 11 vs 23 HALO, whether the drug was given alone or with cisplatin (P=0.001). Gemcitabine-cisplatin tolerability was improved by dosing gemcitabine at 11 HALO and CDDP at 15 HALO (P<0.001). The administration of this schedule to tumour-bearing mice increased median survival three-fold as compared to treatments where both drugs were given simultaneously at 11 or 23 HALO (P=0.02). The optimal schedule would correspond to the delivery of gemcitabine upon awakening and cisplatin near mid-activity in cancer patients.