failure might vary widely. During the past year there has been a lull in the previously very active promotion of primary screening devices. It appears that economic projections for the manufacturers did not materialize. This offers a good opportunity to assess the situation and to suggest where the development might go from here. The first question to be raised requires that we return to the beginning. What is the problem that automation of primary screening is to solve? The often-quoted Wall Street Journal article (Bogdanich W: The pap test misses much cervical cancer through lab’s errors. Nov 2, 1987) called attention to several concerns. There were concerns about the reliability of the reading of a Pap smear, about the possibility of shortages in highly trained personnel and about labor costs. The resulting availability of capital for technology development undoubtedly responded to the need to support the control of cervical cancer, but one has to consider that a prime motivation was the potential for economic gain. There is nothing wrong with that notion in principle, but inevitably it had major consequences. The first consequence was that technical solutions to the problem were explored almost entirely with the medical/economic situation in the United States in mind. A device had to process a slide within a very short time frame so that system throughput would meet revenue projections. Device costs were weighed against amortization schemes, projected unit sales and return on investment capital. There were discussions about acceptable false negative rates and how these might be pegged to revised estimates of false negative rates in current laboratory practice. There were serious and justified concerns about the risk of lawsuits in the United States. One cannot help, though, particularly on the ocAfter 35 years of basic research and development, automated screening devices for cervical cancer went into clinical trials and entered clinical practice. The results have been encouraging and might be called a success in several ways. The United States FDA approved two devices; both proved to perform at a level equal to or better than claimed by the manufacturers. Beyond that the field tests established that it should be possible to design systems that might find approval by the cytopathology community. The devices in the clinical trials were, after all, only first-generation designs. There is no complex technology that performs at the level of its potential until several generations of designs have evolved. As first-generation devices, given the very substantial difficulties of the problem and the circumstances under which their development took place, they performed remarkably well. Eliminating design compromises dictated either by the state of technology at the time of development or by the development climate concomitant with venture capital financing would produce the next generation of primary screening systems that come much closer to a performance level well respected by medical professionals. One might hear the argument that since performance of the first-generation devices was critically assessed by the FDA in the United States, why should the profession call for improvements? It is true that the FDA approved these devices. However, one has to understand that the FDA here merely certified that the devices perform at the level claimed by the manufacturer. If one wanted to be unkind, one might suggest that the industry set its own standard. In practice it would essentially be the laboratory director who decides whether a claimed performance level can be maintained and is acceptable or not. The extent to which such a decision would be based on a thorough understanding