Glaciation has been identified in the rock record through near-field evidence for the presence of glaciers in former landscapes, here called “lithological indicators of glaciation” (LIG). LIG give clues on ice volume, thermal regime, and glaciation timing which, in turn, feed paleoclimatic models. However, LIG involves controversy once its recognition depends on our ability to interpret the rocks from field observation. Uncertainty exists because geological products similar to LIG can be nonglacial, leading to potential misinterpretation of glaciation and, by consequence, false paleoclimatic reconstructions. For instance: “ice-age” diamictite is often generalized as glacial while we know that 1) it is also common in subaquatic landslides and 2) a great part of the “glacial” record includes mass-transport facies. Similarly, we know that lineations may form by drifting ice but soft-sediment striated surfaces are often interpreted as subglacial even when hosted by aquatic deposits. In this contribution I illustrate such controversial issues with two Pennsylvanian case studies from the Itararé Group, a 1.3 km-thick unit in southern Brazil that comprises one of the largest rock records of late Paleozoic glaciations. The first case is a cyclic diamictite-rhythmite succession in which diamictite resulted from downslope collapse of coastal sand/gravel mixed with offshore muds. The second consists of deltaic sandstones with multiple, soft-sediment striations whose characteristics are consistent with keel marks carved by floating ice. Note that both examples were previously interpreted as LIG and as evidence for glacier advance within the basin. LIG definitely exist in the Itararé Group, but when we look at finer stratigraphic scales nonglacial products seem to dominate. Moreover, a search in the literature reveals that nonglacial mass-flow diamictite and keel marks are likely more common than thought in other basins and ages, potentially reducing inferred ice volumes. This kind of ambiguity is the main reason why the extent and longevity of past glaciations have been questioned and shows that reexamination of LIG is essential to our knowledge on past climates.