Background Handgrip strength (HGS) is associated with poor clinical outcomes, including all-cause, non-cardiovascular, and cardiovascular mortalities. The published cut-off points for HGS are mostly based on community populations from Western countries, lacking information on cancer patients from China. The objective of this study was to establish sex-specific cut-off points for Chinese cancer patients and investigate the effect of low HGS on cancer mortality. Methods We did a retrospective cohort study of patients who were diagnosed with malignant cancer from June 2012 to December 2018. HGS was measured using a hand dynamometer in 8257 cancer patients. Optimal stratification was used to solve threshold points. The hazard ratio (HR) of all cancer mortality and cancer-specific mortality was calculated using Cox proportional hazard regression models. Results Among all participants, there were 3902 (47.3%) women and 4355 (52.7%) men. The median age was 58 years old. The cut-off points of HGS to best classify patients with respect to time to mortality were <16.1 kg for women and <22 kg for men. Low HGS was associated with overall cancer mortality in both women and men [HR = 1.339, 95% confidence interval (CI) = 1.170-1.531,P < 0.001; HR = 1.346, 95% CI = 1.176-1.540,P < 0.001, respectively]. For specific cancer types, low HGS was associated with breast cancer (HR = 1.593, 95% CI = 1.230-2.063,P < 0.001) in women, and lung cancer (HR = 1.369, 95% CI = 1.005-1.866,P = 0.047) and colorectal cancer (HR = 1.399, 95% CI = 1.007-1.944,P = 0.045) in men. Conclusions On the basis of our sex-specific cut-off points, low HGS was strongly associated with cancer mortalities. These results indicate the usefulness of HGS measurement in routine clinical practice for improving patient assessments, cancer prognosis, and intervention.