OBJECTIVE: Insulin and the insulin-like growth factor (IGF) system regulate growth and are involved in determining muscle mass, strength and body composition. We hypothesised that IGF-I and IGF-II are associated with improved, and insulin with worse, physical performance in old age. METHODS: Physical performance was measured using the get-up and go timed walk and flamingo balance test at 63-86 years. We examined prospective associations of insulin, IGF-I, IGF-II and IGFBP-3 with physical performance in the UK-based Caerphilly Prospective Study (CaPS; n = 739 men); and cross-sectional insulin, IGF-I, IGF-II, IGFBP-2 and IGFBP-3 in the Boyd Orr cohort (n = 182 men, 223 women). RESULTS: In confounder-adjusted models, there was some evidence in CaPS that a standard deviation (SD) increase in IGF-I was associated with 1.5% faster get-up and go test times (95% CI: -0.2%, 3.2%; p = 0.08), but little association with poor balance, 19 years later. Coefficients in Boyd Orr were in the same direction as CaPS, but consistent with chance. Higher levels of insulin were weakly associated with worse physical performance (CaPS and Boyd Orr combined: get-up and go time = 1.3% slower per SD log-transformed insulin; 95% CI: 0.0%, 2.7%; p = 0.07; OR poor balance 1.13; 95% CI; 0.98, 1.29; p = 0.08), although associations were attenuated after controlling for body mass index (BMI) and co-morbidities. In Boyd Orr, a one SD increase in IGFBP-2 was associated with 2.6% slower get-up and go times (95% CI: 0.4%, 4.8% slower; p = 0.02), but this was only seen when controlling for BMI and co-morbidities. There was no consistent evidence of associations of IGF-II, or IGFBP-3 with physical performance. CONCLUSIONS: There was some evidence that high IGF-I and low insulin levels in middle-age were associated with improved physical performance in old age, but estimates were imprecise. Larger cohorts are required to confirm or refute the findings.