We investigated the age–metallicity relation using a sample of 5691 F- and G-type dwarfs from RAdial Velocity Experiment Data Release 3 (RAVE DR3) by applying several constraints. (i) We selected stars with surface gravities log g(cm s−2) ≥ 3.8 and effective temperatures in the $5310\le T_{\text{eff}}\text{(K)}\le 7300$ range and obtained a dwarf sample. (ii) We plotted the dwarfs in metallicity sub-samples in the $T_{\text{eff}}\text{--}(J-K_s)_0$ plane to compare with the corresponding data of González Hernández & Bonifacio (2009) and identified the ones in agreement. (iii) We fitted the reduced dwarf sample obtained from constraints (i) and (ii) to the Padova isochrones and re-identified those which occupy the plane defined by isochrones with ages t ≤ 13 Gyr. (iv) Finally, we omitted dwarfs with total velocity errors larger than 10.63 km s−1. We estimated the ages using the Bayesian procedure of Jørgensen & Lindegren (2005). The largest age–metallicity slope was found for early F-type dwarfs. We found steeper slopes when we plotted the data as a function of spectral type rather than Galactic population. We noticed a substantial scatter in metallicity distribution at all ages. The metal-rich old dwarfs turned out to be G-type stars which can be interpreted as they migrated from the inner disc or bulge.