BACKGROUND: Consistent evidence has shown excess mortality associated with cold temperature, but some important details of the cold-mortality association (e.g. slope and threshold) have not been adequately investigated and few studies focused on the cold effect in high-altitude areas of developing countries. We attempted to quantify the cold effect on mortality, identify the details, and evaluate effect modification in the distinct subtropical plateau monsoon climate of Yuxi, a high plateau region in southwest China. METHODS: From daily mortality and meteorological data during 2009-2014, we used a quasi-Poisson model combined with a "natural cubic spline-natural cubic spline" distributed lag non-linear model to estimate the temperature-mortality relationship and then a simpler "hockey-stick" model to investigate the cold effect and details. RESULTS: Cold temperature was associated with increased mortality, and the relative risk of cold effect (1st relative to 10th temperature percentile) on non-accidental, cardiovascular, and respiratory mortality for lag 0-21 days was 1.40 (95% confidence interval: 1.19-1.66), 1.61 (1.28-2.02), and 1.13 (0.78-1.64), respectively. A 1 degrees C decrease below a cold threshold of 9.1 degrees C (8th percentile) for lags 0-21 was associated with a 7.35% (3.75-11.09%) increase in non-accidental mortality. The cold-mortality association was not significantly affected by cause-specific mortality, gender, age, marital status, ethnicity, occupation, or previous history of hypertension. CONCLUSIONS: There is an adverse impact of cold on mortality in Yuxi, China, and a temperature of 9.1 degrees C is an important cut-off for cold-related mortality for residents.