The various sources of random error, occurring in laser microprobe analysis using cross-excitation have been examined. It is shown that the combination of laser sampling and cross-excitation constitutes the major source of random error. In an attempt to reduce this error, the spectral line intensity has been related directly to a reference line, to the crater depth, and to the crater volume. However, introduction of these parameters did not improve the analytical results. The crater depth formula proposed by Bar-Isaac and Korn appeared to be of little analytical use, not only because all numerical values of the physical constants involved have to be exactly known, but also because each laser mode gives rise to different removal processes. A reproducibility of the spectral line intensity up to 16% is met. Moreover, due to the small slopes of the analytical curves, the precision of the analysis is rather poor. The use of the laser microprobe (employing cross-excitation) in quantitative analyses is therefore to be dissuaded. © 1979.