Abstract. Security API analysis typically only considers a subset of an API's functions, with results bounded by the number of function calls. Furthermore, attacks involving partial leakage of sensitive information are usually not covered. Type-based static analysis has the potential to alleviate these shortcomings. To that end, we present a type system for secure information flow based upon the one of Volpano, Smith and Irvine [1], extended with types for cryptographic keys and ciphertext similar to those in Sumii and Pierce [2]. In contrast to some other type systems, the encryption and decryption of keys does not require special treatment. We show that a well-typed sequence of commands is non-interferent, based upon a definition of indistinguishability where, in certain circumstances, the adversary can distinguish between ciphertexts that correspond to encrypted public data.