BACKGROUND: Short-term starvation prior to chemotherapy administration protects mice against toxicity. We undertook dose-escalation of fasting prior to platinum-based chemotherapy to determine safety and feasibility in cancer patients. METHODS: 3 cohorts fasted before chemotherapy for 24, 48 and 72 h (divided as 48 pre-chemo and 24 post-chemo) and recorded all calories consumed. Feasibility was defined as >/= 3/6 subjects in each cohort consuming </= 200 kcal per 24 h during the fast period without excess toxicity. Oxidative stress was evaluated in leukocytes using the COMET assay. Insulin, glucose, ketones, insulin-like growth factor-1 (IGF-1) and IGF binding proteins (IGFBPs) were measured as biomarkers of the fasting state. RESULTS: The median age of our 20 subjects was 61, and 85 % were women. Feasibility criteria were met. Fasting-related toxicities were limited to </= grade 2, most commonly fatigue, headache, and dizziness. The COMET assay indicated reduced DNA damage in leukocytes from subjects who fasted for >/=48 h (p = 0.08). There was a non-significant trend toward less grade 3 or 4 neutropenia in the 48 and 72 h cohorts compared to 24 h cohort (p = 0.17). IGF-1 levels decreased by 30, 33 and 8 % in the 24, 48 and 72 h fasting cohorts respectively after the first fasting period. CONCLUSION: Fasting for 72 h around chemotherapy administration is safe and feasible for cancer patients. Biomarkers such as IGF-1 may facilitate assessment of differences in chemotherapy toxicity in subgroups achieving the physiologic fasting state. An onging randomized trial is studying the effect of 72 h of fasting. TRIAL REGISTRATION: NCT00936364 , registered propectively on July 9, 2009.