Context. The age–velocity dispersion relation is an important tool to understand the evolution of the disc of the Andromeda galaxy (M 31) in comparison with the Milky Way. Aims. We use planetary nebulae (PNe) to obtain the age–velocity dispersion relation in different radial bins of the M 31 disc. Methods. We separate the observed PNe sample based on their extinction values into two distinct age populations in the M 31 disc. The observed velocities of our high- and low-extinction PNe, which correspond to higher- and lower-mass progenitors, respectively, are fitted in de-projected elliptical bins to obtain their rotational velocities, V_ϕ, and corresponding dispersions, σ_ϕ. We assign ages to the two PN populations by comparing central-star properties of an archival sub-sample of PNe, that have models fitted to their observed spectral features, to stellar evolution tracks. Results. For the high- and low-extinction PNe, we find ages of ∼2.5 and ∼4.5 Gyr, respectively, with distinct kinematics beyond a deprojected radius R_GC = 14 kpc. At R_GC = 17–20 kpc, which is the equivalent distance in disc scale lengths of the Sun in the Milky Way disc, we obtain σ_{ϕ, 2.5 Gyr} = 61 ± 14 km s⁻¹ and σ_{ϕ, 4.5 Gyr} = 101 ± 13 km s⁻¹. The age–velocity dispersion relation for the M 31 disc is obtained in two radial bins, R_GC = 14–17 and 17–20 kpc. Conclusions. The high- and low-extinction PNe are associated with the young thin and old thicker disc of M 31, respectively, whose velocity dispersion values increase with age. These values are almost twice and three times that of the Milky Way disc stellar population of corresponding ages, respectively. From comparison with simulations of merging galaxies, we find that the age–velocity dispersion relation in the M 31 disc measured using PNe is indicative of a single major merger that occurred 2.5–4.5 Gyr ago with an estimated merger mass ratio ≈1:5.