BACKGROUND: Significant socio-economic disparities have been reported in survival from colorectal cancer in a number of countries, which remain largely unexplained. We assessed whether possible differences in access to treatment among socio-economic groups may contribute to those disparities, using a population-based approach. METHODS: We retrospectively studied 71 917 records of colorectal cancer patients, diagnosed between 1997 and 2000, linked to area-level socio-economic information (Townsend index), from three cancer registries in UK. Access to treatment was measured as a function of delay in receipt of treatment. We assessed socio-economic differences in access through logistic regression models. Based on relative survival < or =3 years after diagnosis, we estimated excess hazard ratios (EHRs) of death for different socio-economic groups. RESULTS: Compared with more affluent patients, deprived patients had poorer survival [EHR = 1.20; 95% confidence interval (CI) 1.16-1.25], were less likely to receive any treatment within 6 months [odds ratio (OR) = 0.87, 95% CI 0.82-0.92] and, if treated, were more likely to receive late treatment. No disparities in survival were detected among patients receiving treatment within 1 month from diagnosis. Disparities existed among patients receiving later or no treatment (EHR = 1.30; 95% CI 1.22-1.39), and persisted after adjustment for age and stage at diagnosis (EHR = 1.15; 95% CI 1.08-1.24). CONCLUSIONS: Tumour stage helped explain socio-economic disparities in colorectal cancer survival. Disparities were also greatly attenuated among patients receiving early treatment. Aspects other than those captured by our measure of access, such as quality of care and patient preferences in relation to treatment, might contribute to a fuller explanation.