Higher body mass index (BMI) is associated with incident colorectal cancer but not consistently with colorectal cancer survival. Whether weight gain or loss is associated with colorectal cancer survival is largely unknown. We identified 2,781 patients from Kaiser Permanente Northern California diagnosed with stages I-III colorectal cancer between 2006 and 2011 with weight and height measurements within 3 months of diagnosis and approximately 18 months after diagnosis. We evaluated associations between weight change and colorectal cancer-specific and overall mortality, adjusted for sociodemographics, disease severity, and treatment. After completion of treatment and recovery from stage I-III colorectal cancer, loss of at least 10% of baseline weight was associated with significantly worse colorectal cancer-specific mortality (HR 3.20; 95% confidence interval [CI], 2.33-4.39; Ptrend< 0.0001) and overall mortality (HR 3.27; 95% CI, 2.56-4.18; Ptrend< 0.0001). For every 5% loss of baseline weight, there was a 41% increased risk of colorectal cancer-specific mortality (95% CI, 29%-56%). Weight gain was not significantly associated with colorectal cancer-specific mortality (Ptrend= 0.54) or overall mortality (Ptrend= 0.27). The associations were largely unchanged after restricting analyses to exclude patients who died within 6 months and 12 months of the second weight measurement. No significant interactions were demonstrated for weight loss or gain by gender, stage, primary tumor location, or baseline BMI. Weight loss after diagnosis was associated with worse colorectal cancer-specific mortality and overall mortality. Reverse causation does not appear to explain our findings. Understanding mechanistic underpinnings for the association of weight to worse mortality is important to improving patient outcomes. Cancer Epidemiol Biomarkers Prev; 26(1); 30-37. ©2016 AACR SEE ALL THE ARTICLES IN THIS CEBP FOCUS SECTION, "THE OBESITY PARADOX IN CANCER EVIDENCE AND NEW DIRECTIONS".