Objective: The purpose of this study was to establish a depth-control method in enamel-cavity ablation by optimizing the timing of the focal-plane-normal stepping and the single-step size of a three axis, numerically controlled picosecond laser. Background data: Although it has been proposed that picosecond lasers may be used to ablate dental hard tissue, the viability of such a depth-control method in enamel-cavity ablation remains uncertain. Methods: Forty-two enamel slices with approximately level surfaces were prepared and subjected to two-dimensional ablation by a picosecond laser. The additive-pulse layer, n, was set to 5,10,15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70. A three-dimensional microscope was then used to measure the ablation depth, d, to obtain a quantitative function relating n and d. Six enamel slices were then subjected to three dimensional ablation to produce 10 cavities, respectively, with additive-pulse layer and single-step size set to corresponding values. The difference between the theoretical and measured values was calculated for both the cavity depth and the ablation depth of a single step. These were used to determine minimum-difference values for both the additive-pulse layer (n) and single-step size (d). Results: When the additive-pulse layer and the single-step size were set 5 and 45, respectively, the depth error had a minimum of 2.25 lm, and 450 lm deep enamel cavities were produced. Conclusions: When performing three-dimensional ablating of enamel with a picosecond laser, adjusting the timing of the focal-plane-normal stepping and the single-step size allows for the control of ablation-depth error to the order of micrometers.