The accuracy of a two-sample zero-crossing detection method is assessed by analytical uncertainty propagation and is verified in numerical simulations. Approximated expressions are given to evaluate uncertainty components due to linear interpolation, quantization, white noise and time jitter. The combined standard uncertainty of a detected zero phase is expressed as a function of the signal frequency, power of external noise and parameters of the digitizer used. The evaluation of the uncertainty of the measurand, derived from several detected zero-crossings, is illustrated with applications in frequency, displacement and free-fall acceleration measurements. The reported results can be used for uncertainty analysis and parameter optimization of a measurement system or a procedure, involving processing of sinusoidal signals.