BACKGROUND: No study has been conducted on the scientific quality of randomized controlled trials (RCTs) in the cancer field. Our objective was to determine whether adherence to the Consolidated Standards for Reporting Trials (CONSORT) statement is associated with scientific properties of RCT reports from two leading cancer journals. METHODS: We conducted an observational study of RCTs published between 2002 and 2004 in two leading cancer journals that did not endorse the CONSORT statement during that period. We determined the adherence rates with confidence intervals of 33 RCTs according to the 19 methodological items of the CONSORT statement. Each RCT was blindly assessed by three independent evaluators; then the evaluators examined all judgments sequentially and obtained a consensus regarding each methodological item of the CONSORT statement. RESULTS: The average adherence of these 33 RCTs to the 19 methodological items of the CONSORT statement was 79.3% (95% CI, 75.3-83.4%). Most descriptors from the checklist were determined to be methodologically adequate except sequence generation (56.1%; 95% CI, 40.9-71.3%), allocation concealment (27.3%; 95% CI, 13.2-41.4%), implementation (7.6%; 95% CI, 0.0-15.4%), blinding (30.3%; 95% CI, 14.4-46.3%) and sample size (74.2%; 95% CI, 59.5-89.0%). Of all CONSORT checklist items, randomization implementation was the most often omitted. CONCLUSION: Some key methodological items of the CONSORT statement seem poorly addressed in RCTs from these leading cancer journals. Thus researchers should be urged to conform to the CONSORT statement when reporting on RCTs, and the poorly addressed items of the CONSORT statement should be reevaluated for RCTs already reported.