Inevitable errors in X-probe measurement are investigated in simulated Gaussian velocity fields generated on a digital computer. Two types of X arrangement are studied: one is a conventional X-probe comprised of two inclined hot wires, and the other consists of inclined and normal hot wires (modified X-probe). The computer simulation results under typical wall-turbulence conditions indicate that the most significant factor producing measurement errors, in both high and low intensity turbulence, is the difference between the instantaneous velocities sensed by the two wires, associated with wire separation. In particular, the measurement accuracy of the modified X-probe depends greatly on the effect of wire separation, and at worst, data obtained with such a probe result in fallacious information. To confirm the error analysis, turbulent quantities in a pipe flow were measured using these X-probes. The experimental results are consistent with the results of the error analysis. Also, based on the present analysis, simple formulae have been developed to estimate inherent errors in X-probe measurements.