Measures of lexical frequency presuppose the existence of corpora, but true machine-readable corpora of sign languages (SLs) are only now being created. Lexical frequency ratings for SLs are needed because there has been a heavy reliance on the interpretation of results of psycholinguistic and neurolinguistic experiments in the SL research literature; yet, these experiments have been conducted without the benefit of such measures. In addition, measures of lexical frequency can also guide SL teachers by identifying which signs would be prioritized in early language instruction. I begin by a discussion of lexicalization and sign types in order to explain what constitutes a lexical sign in SLs. I then present the annotation method and results. In the discussion, I raise the potential limitations of previous studies of lexical frequency in terms of the discrimination of lexical signs from other kinds of signs, consistent lemma glossing, part of speech tagging, and the systematic treatment of depicting signs. I conclude in cautioning that descriptions of SL grammars that do not accommodate typical mixtures and sequences of signs as shown in data are likely to be unreliable.