The spontaneous use of analogies in problem-solving occurs when a subject first spontaneously shifts his attention to a situation (B) which differs in some significant way from an original problem situation (A), and then tries to apply findings from B to A. This paper describes research on the process with 10 scientifically trained subjects (advanced doctoral students and professors in technical fields) who were interviewed (most were videotaped) on a variety of problems. Four processes were found to be fundamental in making an inference by analogy, indicating that many successful solutions by analogy are not "instant solutions." Two major processes involved in confirming analogy relations were identified: matching key features or relationships and forming a bridging analogy. Analysis of transcripts led to an hypothesis that at least three types of analogy generation mechanisms exist: generation via an abstract principle, generative transformations, and associative leaps, the latter two mechanisms being the primary methods observed so far. Since many solutions by analogy are not "instant solutions," but a more extended process of conjuncture and testing, it is suggested that some of these processes are learnable and that developing students' abilities to use generative transformations, leaps, and bridges may be possible and desirable. (JN)