We sought to examine the impact of hospital surgical volume on the number of nodes harvested and survival in colorectal cancer (CRC). Between January 1994 and December 2004, a total of 8567 patients with T1, 2, 3, and 4 primary tumors and N0, N1, or N2 disease were studied. Hospitals were stratified into very low volume (VLV) (<33 cases/year), low volume (LV) (33-56 cases/year), and medium volume (MV) (57-84 cases/year). Surgery for CRC was performed most commonly at VLV hospitals: 3488 (40.7%) VLV centers versus 2359 (27.5%) LV centers versus 2720 (31.7%) MV centers. The mean number of nodes retrieved for VLV centers was 8.6, for LV centers 9.4, and MV centers 10.2 (P < 0.0002). Actuarial 5-year survival for VLV centers was 71.4 per cent, for LV centers 75.6 per cent, and for MV 77.0 per cent (P < 0.00001). By Cox proportional hazards analysis, hospital volumes (P < 0.0011) and the number of lymph nodes harvested (P < 0.0034) remain significant predictors of disease specific survival. The number of nodes retrieved is impacted by hospital volumes. Hospital volumes impact survival in CRC. These findings cannot be attributed solely to improved staging due to increased node retrieval in VLV, LV, and MV hospitals.