Some general points, since the question is somewhat vague:
In order to get good switching speed with a MOSFET, you need to actively drive the gate, and actively discharge the gate. This can be accomplished with a discrete two transistor totem-pole gate driver, or (more commonly these days) with monolithic gate driver ICs that have amps of instantaneous sink/source capability.
One thing that some people don't consider is the internal gate resistance of the MOSFET, which manifests itself in series with the interal gate-to-source and gate-to-drain capacitances. Consider this simplified model (courtesy Vishay):

\$R_g\$ is one of the main reasons you need a stiff driver, since it limits the current available to charge and discharge \$C_{gs}\$. (\$C_{gs}\$ itself is another reason.)
\$C_{gd}\$ is also known as the 'Miller' capacitance, is quite nonlinear and also plays a role in gate turn-on and turn-off, since the gate needs to overcome both of these capacitances to get the device solidly 'on'.

Another area that can cause slower than necessary turn-off (with N-channel MOSFETs) is if the applied gate voltage is much higher than the voltage required to turn the device on. If you have a logic-level MOSFET and need only 5V to fully enhance the channel, and apply 12V to the gate, that extra 7 volts doesn't get you much except extra \$C_{gs}\$ charge to remove when you're trying to turn the device off. If you're using a gate driver that can sink or source amps of current, this extra voltage isn't really a concern. If you're passively discharging the gate, it can significantly affect the commutation time.