Repeated surveys of selected lines from five trilateration networks along the San Andreas fault in southern California have been used to deduce the 1973--1984 strain accumulation records at five localities. The secular rate of engineering shear strain accumulation is about 0.3 μrad/yr with the plane of maximum shear parallel to the local strike of the San Andreas fault. The secular rate of accumulation of areal dilatation is negligible. The data were examined to detect evidence for fluctuations in the rate of strain accumulation. For this examination, 19 lines were removed from the data set: four because they exhibited an obvious coseismic offset and 15 others because they contained at leaast one very anomalous measurement that could resonably be attributed to a survey blunder. (The incidence of such blunders appears to be one in every 75 measurements.) The remaining data consist of 104 lines with an average of 10 measurements each. Although the strain accumulation plots for the five networks may exhibit marginally significant temporal fluctuations, we are not convinced that those fluctuations are greater than could be attributed to survey error. In particular, we are unable to demonstrate that the 1973--1979 southern California strain anomaly reported by Savage and others is real. Given the uncertainty in the random and systematic errors in measurement, the strain measurements in southern California are marginally consistent with linear-in-time strain accumulation. The strain accumulation plots for the Salton network clearly established that, unlike the deformation reported after the 1940 Imperial Valley earthquake, no acceleration in the shear strain rate has yet been observed following the 1979 Imperial Valley earthquake.