The application of a small computer to the individual scientist's bibliographic problems is described. Journal references are entered into the computer, which then types back a reprint request (to be sent in a window envelope) and three file cards. The initial typing of the reference is thus reproduced (with typing errors corrected) four times by the computer, as well as being saved permanently on magnetic tape for future use. While everyday needs are handled by the file card system (using subject classifications created by the user), special searches can be made of the cumulative list of references on the computer tape. No subject classifications are included in the computer, so searches are based upon remembered fragments of the reference (What was that article on calcium in J. Physiol. last year'? Search for CALCIUM and J. PHYSIOL,). Keywords may be appended to the citation itself at the time of original entry if desired, and included in the searched material. Because no scientific decisions are involved in the citation entry into the computer, it can be a secretarial job, with the user making scientific decisions when filing the 3 X 5 file cards (using his own personal system) and selecting topics for searches of the computerized cumulative reference list. Search strategy need not focus upon machine extraction of a single citation among thousands; rather, it attempts to define a subset which may be further sorted manually (by pressing the rubout key when an undesirable refercnce is dEp!~yed on the computer scope prior to typing). The system currently operates on a LINC-8 computer, but may be readily adapted to any small computer having a tape and a good text-editor.