Many studies have previously reported that the recognition of a stem target (e.g., "teach") is facilitated by the prior masked presentation of a prime consisting of a derived form of it (e.g., "teacher"). We conducted two lexical decision experiments to investigate masked morphological priming in Spanish. Experiment 1 showed that equal magnitudes of masked stem-target priming are obtained for both morphologically complex word primes (e.g., "doloroso-DOLOR" ["painful-PAIN"]) and morphologically complex nonword primes that included letter transpositions within the stem (e.g., "dlooroso-DOLOR"). Experiment 2 used morphologically complex nonword primes comprising lexically illegal combinations of stems and suffixes (e.g., "total + ito" ["a little total"]). Priming was obtained for morphologically related nonword primes (e.g., "totalito-TOTAL"), but not for nonword primes that included letter transpositions within the pseudostem (e.g., "ttoalito-TOTAL"). Our data suggest that morphoorthographic parsing mechanisms benefit from semantic constraints at early stages in the reading system, which we discuss in the context of current morphological processing accounts.