Using perturbative QCD approach, charmless $B_{u,d,s}\to VT$ decays are predicted. Their branching ratios, polarization fractions and direct CP violations are calculated. Within this approach the polarization fractions and the branching ratios of $B\to \phi(K_2^{*-}, \bar K_2^{*0})$ agree with the observed experimental data. However, the branching ratios of $B\to \omega(K_2^{*-}, \bar K_2^{*0})$ cannot be explained, where the polarization fractions can be accommodated. The tree dominated channels with a vector meson emitted have longitudinal polarization fraction of $90\%$, while the penguin dominating ones have subtle polarization fractions. Fortunately, most branching ratios of $B_{u,d}$ decays are of the order $10^{-6}$, which would be straight forward for experimental observations. For the $B_s$ decays, the branching ratios can reach the order of $10^{-6}$ in a vector emitted tree dominated decays. However, in penguin dominated decays those are of order of $10^{-7}$ which need more experimental data to be observed.