The　basic　problems　of　autofocus　in　digital　holographic　microscopy,　such　as　numerical　reconstruction　algorithms,　zero-order　spectrum　filtering　of　hologram　and　focusing　evaluation　functions,　are　studied　by　theoretical　analysis　and　experimental　verification.　The　results　show　that　Fresnel　transform　algorithm　can　be　used　in　digital　holographic　autofocus;zero-order　spectrum　filtering　of　hologram　leads　to　unsuccessful　focusing　evaluation　which　is　based　on　Fresnel　transform　algorithm;　gray　variance　function,　weighted　Fourier　spectrum　logarithm　function　and　standard　deviation　correlation　function　can　be　effectively　applied　in　the　autofocus　of　digital　holographic　microscopy.　Among　these　functions,　the　Fourier　spectrum　logarithm　function　is　the　best　focusing　evaluation　function　due　to　its　lowest　time　cost.　Moreover,　the　focusing　evaluation　time　can　be　shortened　dramatically　by　using　the　part　of　the　reconstructed　image　field　as　the　focusing　evaluation　area.