The forced oscillation technique (FOT) allows the measurement of respiratory resistance (Rrs) and reactance (Xrs) and their associated coherence (gamma2). To avoid unreliable data, it is usual to reject Rrs and Xrs measurements with a gamma2 <0.95. This procedure makes it difficult to obtain acceptable data at the lowest frequencies of interest. The aim of this study was to derive expressions to compute the random error of Rrs and Xrs from gamma2 and the number (N) of data blocks involved in a FOT measurement. To this end, we developed theoretical equations for the variances and covariances of the pressure and flow auto- and cross-spectra used to compute Rrs and Xrs. Random errors of Rrs and Xrs were found to depend on the values of Rrs and Xrs, and to be proportional to ((1-gamma2)/(2 x N x gamma2))1/2. Reliable Rrs and Xrs data can be obtained in measurements with low gamma2 by enlarging the data recording (i.e. N). Therefore, the error equations derived may be useful to extend the frequency band of the forced oscillation technique to frequencies lower than usual, characterized by low coherence.