The analysis of quartz grain orientation in a set of 261 thin sections obtained from the Thorold and Grimsby Sandstones (Silurian, S.W. Ontario and New York State) indicates that: (1) The number of measurements needed for a petrographic analysis of grain orientation varies from 50-100 in thin sections cut perpendicular to the depositional surface, to 150-250 in thin sections cut parallel to the depositional surface. (2) Preferred grain imbrications are encountered in the vast majority of the samples examined, regardless of the direction of cut of the vertical thin sections with respect to the vector mean of the grain orientation measured parallel to the depositional surface (H). However, the vertical thin sections cut parallel to the vector mean of the H-grain orientation show a tendency to Lave an average imbrication value significantly different from zero, hence usable for paleocurrent determinations. (3) Non-significant grain orientation distributions are useful in focusing attention on sedimentological characteristics needed for paleoenvironmental reconstructions. (4) Erroneous results are obtained by considering the principal bedding surface as the reference plane for the grain imbrications of apparently massive beds when these are in fact, cross-bedded units. The grain imbrication populations measured in apparently massive beds are a mixture of grain imbrications and of H-grain orientations. The vector means of such a grain imbrication populations cannot be utilized for paleocurrent determinations. The vector means of the H-grain orientation of the massive beds can, instead, be utilized as valid line of movement data although, in this case, the possibility exists of accepting a non-preferred grain orientation as a significant one.