A high-resolution (1.25 m) LES simulation of the nocturnal cloud-topped marine boundary layer is used to evaluate random error as a function of continuous track length L for virtual aircraft measurements of turbulent fluxes of sensible heat, latent heat, and horizontal momentum. Results are compared with the theoretically derived formula of Lenschow and Stankov (1986). In support of these comparisons , we also evaluate and document the relevant integral length scales and correlations and show that for heights up to approximately 100 m (z/z i = 0.12), the length scales are accurately predicted by empirical expressions of the form I f = Az^b. The Lenschow and Stankov expression is found to be remarkably accurate at predicting the random error for shorter flight tracks, but our empirically determined errors decay more rapidly with L than the L^−1/2 relationship predicted from theory. Consistent with earlier findings, required track lengths to obtain useful precision increase sharply with altitude.