Source monitoring involves attributing information to one of several sources. Schemas are known to influence source-monitoring processes, with enhanced memory for schematically unexpected sources (inconsistency effect) and biased schema-consistent source guessing. The authors investigated whether this guessing bias reflects a compensatory guessing strategy based on metacognitive awareness of the inconsistency effect, or reflects other strategies as proposed by the probability-matching account. To determine people's awareness of the inconsistency effect, the authors investigated metamemory predictions in a source-monitoring task. In four experiments, participants studied object word items that were presented with one of two scene labels as sources. Items were either presented with their schematically expected source (e.g., kitchen-- oven) or with their schematically unexpected source (e.g., kitchen-- toothpaste). In Experiments 1 and 2, participants predicted their item memory and their source memory after each source--item presentation. In Experiment 1, people incorrectly predicted both their item memory and, even more so, their source memory to be better for expected than for unexpected source--item pairs. In Experiment 2, this effect replicated with different types of judgment probes. Crucially, item-wise memory predictions did not predict source guessing. In Experiment 3, metacognitive awareness of the inconsistency effect on source memory changed during the test phase. However, metamemory convictions never predicted source guessing. In Experiment 4, the authors manipulated participants' convictions concerning the impact of schematic expectations on source memory. These convictions also did not predict source guessing. Thus, the results show that schema-consistent source guessing does not reflect a compensatory strategy.