We have estimated the frequency of release factor dependent events in which a sense codon is mistakenly translated as a stop codon. We refer to this event as a "false stop". In order to facilitate the measurement of false stop freqeuncies we have used a plasmid expression system to increase individually the cellular levels of release factor (RF) I a and of release factor (RF) 2. We were then able to measure the loss of translational processivity with the aid of a lacZ processivity assay at different concentrations of the release factors. We find that a 30- to 40-fold increase of the RF1 concentration reduces lacZ processivity from 0.6 to 0.3. Assuming that the processivity loss is due only to false stops and that the RF1 overproduction data can be extrapolated back linearly to the normal RF1 concentration in the cell, this corresponds to a false stop frequency close to 10(-5) per codon in the presence of normal amounts of RF1. Furthermore, a threefold increase of the RF2 concentrations had no measurable effect on the processivity of the lacZ gene. Our data suggest that false stops are relatively infrequent compared to the incidence of other translation errors.