Survival estimates of patients registered by population-based cancer registries reflect the average prognosis from a given cancer as they are based on unselected patients with a wide range of natural histories and treatment patterns. In this paper, we report the survival experience of colorectal cancer patients in Mumbai (Bombay), India. Follow-up information on 1642 colorectal cancer patients registered by the Bombay Population-based Cancer Registry for the period 1987-1991 was obtained by matching with death certificates from the Bombay vital statistics registration system, postal/telephone enquiries, home visits and scrutiny of medical records. Cumulative observed and relative survival proportions were calculated by Hakulinen's method. For comparison of results with other populations, age-standardised relative survival (ASRS) was calculated by directly standardising age-specific relative survival to the specific age distributions of the world standard cancer patient population in 1985. The log-rank test was used to identify the potential prognostic variables which were introduced step-wise into a Cox regression model to identify the independent predictors of survival. The 5-year relative survival was 36.6% for colon and 42.2% for rectal cancer. Age, site of cancer and clinical stage of disease emerged as independent predictors of survival. Age-specific 5-year relative survival declined with advancing age. Survival at 5 years was 61.2% for localised colon cancer; 31.9% for regional and 9.0% for distant metastatic disease. These were 65.7, 25.6 and 4.3%, respectively for rectal cancers. Comparison of the results with other populations revealed significant variations, which seem to be related to differences in detection and treatment. The prognosis from colorectal cancer in Mumbai and developing countries, may be further improved through early detection linked with treatment.