The urgency of the work is caused by the fact that the power spectrum as well as the correlation function, is one of the most important characteristics of the second order of stochastic process. The spectrum shows what kind of harmonics prevails in the process, its structure; allows estimating spectral composition of the studied useful signals and noises. It is possible to synthesize (recover) signal, as well as to construct linear, including optimal filters, to obtain error estimates of linear filtering by the spectra. The main aim of the study is to evaluate the power spectrum of a stationary random process as a first-order spline under the following measurement schemes: one measurement is made every moment, a random number of measurements is made every moment. Study of statistical characteristics of the estimates. The methods used in the study: the methods of probability theory and mathematical statistics are used for calculation. The results: The authors obtained an unbiased estimate of the power spectrum as a first-order spline in two schemes of measurement: one measurement is made every moment, a random number of measurements is made every moment. It is shown that the variance of estimates behaves asymptotically like 1 / T, where Т is the time of observation.