In this paper, a criterion for a fast automatic approach method in conventional atomic force microscope is introduced. There are currently two approach methods: automatic and semiautomatic methods. However, neither of them provides a high approach speed to enable the avoidance of possible damage to tips or samples. Industrial atomic force microscope requires a high approach speed and good repeatability for inspecting a large volume. Recently, a rapid automatic engagement method was reported to improve the approach speed. However, there was no information on how to determine the safe distance. This lack of information increases the chance for damage to occur in calibrating optimal approach speed. In this paper, we show that the proposed criterion can be used for decision making in determining mode transitions from fast motion to slow motion. The criterion is calculated based on the average intensity of a Gaussian laser beam. The tip-sample distance where the average intensity becomes the maximum value is used for the criterion. We explain the effects of the beam spot size and the window size on the average intensity. From experimental results with an optical head used in a commercial atomic force microscope, we observed that the mean and standard deviation (of the distance at which intensity is the maximum for the 25 experiments) are 194.0 and 15.0 microm, respectively, for a rectangular cantilever (or 224.8 and 12.6 microm for a triangular cantilever). Numerical simulation and experimental results are in good agreement.