15N relaxation rates of amide moieties provide insight both into global as well as local backbone dynamics of peptides and proteins. As the differences in the relaxation rates in general are small, their accurate determination is of prime importance. One potential source of error is fast amide exchange. It is well known that in its presence the effects of saturation transfer and H/D exchange may result in erroneous apparent relaxation rates R1 and R2. Here, the extent of these errors is rigorously examined. Theoretical considerations reveal that even when saturation effects are absent, H/D exchange will easily result in significant deviations from the true values. In particular overestimations of up to 10% in R1 and up to 5% in R2 are observed. Alternative scheme for fitting the relaxation data to the corresponding exponentials is presented that in the best cases not only delivers more accurate relaxation rates but also allows extracting estimates for the exchange rates. The theoretical computations were tested and verified for the case of ubiquitin.