Not only the maximum attainable linear density and the efficiency of a modulation code, but also the complete error picture of the code should be considered when optimizing a design for a high performance disk drive. Operating reliably at a particular density and data rate requires that the sum of all timing errors is less than the data window. These timing errors are noise jitter, errors due to the write process, and electronic errors. Disk capacity as a func-tion of data rate is shown to be a useful measure of the overall effectiveness of a code. In a first step, the analog signals of critical patterns of different codes were digitized and their associated noise levels measured. In an error picture simulation, the influence of filter and differentiator on the waveform was observed, and the signal to noise (S/N) ratio and the noise jitter were calculated. Timing errors due to the write process and electronics were estimated. The effects of rotation speed and data rate were simulated. On condition that all timing errors fit exactly into the data window, the maximum attainable data rate was determined. In a second step, the data rate and data density were verified experimentally. The data density versus data rate curves were determined for the EFM code (pulse edge detection), the RLL 2.7 code (pulse peak detection) and the 4/15 code (differential detection).