Synthetic colours have been calculated for the Washington photometry system from a grid of synthetic spectra. The grid of spectra covers a range in T(eff), log g and abundance appropriate for Population I and II dwarfs and giants. We also examine the effects of C and N abundance variations on the colours of the giants. We compare the model colours with observations of field dwarfs and giants compiled from the literature, as well as published cluster giant branch photometry. We find that our dwarf models match the observed disc dwarf sequence very well in the (C - M) versus (T1 - T2) and (M - T1) versus (T1 - T2) colour planes, while our solar-abundance giant branch grid and isochrones are a good fit to the disc giant sequence. Our metal-poor isochrones are systematically too red when compared with globular cluster giant branches; this may be due to the parameters chosen for the computation of the isochrones. A comparison of the effects of abundance and gravity on the indices (T1 - T2) and (M - T2), used for estimating T(eff), suggests that the latter should be the index of choice. Neither (C - M) nor (M - T1) is found to be particularly sensitive to variations in [C/Fe] or [N/Fe] if the abundances are anticorrelated. At solar abundance, however, (C - M), (M - T1) and (T1 - T2) are strongly affected if the abundances are not anticorrelated. We confirm empirical suggestions that (C - M) and (C - T1) are the best abundance indicators. The model colours for the gravity indicator (M - 51) match the observed sequences very well. This index is useful for estimating both gravity and abundance. It may be particularly useful for estimating the abundance of highly reddened open and globular clusters.