Temperature is arguably the most significant single determinant of water quality but data are collected at comparatively few locations and are rarely subjected to systematic analysis. This paper details two different methods suitable for water managers engaged in predicting river temperatures on an operational basis over a wide range of time scales. Firstly, simple linear regression equations are presented for estimating annual and monthly river temperatures to an acceptable level of accuracy using air temperatures only. Secondly, it is shown how an empirical equation can be fitted to short term observations of river temperature in order to represent the characteristic seasonal and diurnal cycles.