Epidemiological studies have quantified the association between ambient temperature and diarrhoea. However, to our knowledge, no study has quantified the temperature association for severe diarrhoea cases. In this study, we quantified the association between mean temperature and two severe diarrhoea outcomes, which were mortality and hospital admissions accompanied with dehydration and/or co-morbidities. Using a 12-year dataset of three urban districts of the National Capital Region, Philippines, we modelled the non-linear association between weekly temperatures and weekly severe diarrhoea cases using a two-stage time series analysis. We computed the relative risks at the 95th (30.4 °C) and 5th percentiles (25.8 °C) of temperatures using minimum risk temperatures (MRTs) as the reference to quantify the association with high- and low-temperatures, respectively. The shapes of the cumulative associations were generally J-shaped with greater associations towards high temperatures. Mortality risks were found to increase by 53.3% [95% confidence interval (CI): 29.4%; 81.7%)] at 95th percentile of weekly mean temperatures compared with the MRT (28.2 °C). Similarly, the risk of hospitalised severe diarrhoea increased by 27.1% (95% CI: 0.7%; 60.4%) at 95th percentile in mean weekly temperatures compared with the MRT (28.6 °C). With the increased risk of severe diarrhoea cases under high ambient temperature, there may be a need to strengthen primary healthcare services and sustain the improvements made in water, sanitation, and hygiene, particularly in poor communities.