Background: Mental disorders have been found to be positively associated with temperature in cool to cold climatic regions but the association in warmer regions is unclear. This study presented the short-term association between temperatures and mental disorder hospitalizations in a subtropical city with a mean annual temperature over 21 °C. Methods: Using Poisson-generalized additive models and distributed-lagged nonlinear models, daily mental disorder hospitalizations between 2002 and 2011 in Hong Kong were regressed on daily mean temperature, relative humidity, and air pollutants, adjusted for seasonal trend, long-term trend, day-of-week, and holiday. Analyses were stratified by disease class, gender and age-group. Results: 44,600 admissions were included in the analysis. Temperature was positively associated with overall mental-disorder hospitalizations (cumulative relative risk at 28 °C vs. 19.4 °C (interquartile range, lag 0–2 days) = 1.09 (95% confidence interval 1.03, 1.15)), with the strongest effect among the elderly (≥75 years old). Transient mental disorders due to conditions classified elsewhere and episodic mood disorders also showed strong positive associations with temperature. Conclusion: This study found a positive temperature–mental-disorder admissions association in a warm subtropical region and the association was most prominent among older people. With the dual effect of global warming and an aging population, targeted strategies should be designed to lower the disease burden.