The sudden rush to bring to market and hype computer-aided software engineering tools must be leaving the casual observer with the impression that the large IBM mainframe sites that are the most coveted customers in the computer business are panting to reduce their applications development costs by letting the computer take the strain. But a report from Dun & Bradstreet’s Focus Research Systems in West Hartford, Connecticut seems to pour a cold douche over that idea. It reckons that the number of IBM and compatible mainframe computer sites using computer-aided software engineering tools for new development will double in 1988, which sounds enticing, except that the currently only 4.1% of IBM and compatible sites across the US are currently using software engineering tools at present. It finds that over the next 12 months another 4.2% plan to add a product. Computer sites with longer application development backlogs are the biggest initial users of the technology, the study found. Software engineering tools are used at 2.9% of those sites with application backlogs of zero to six months, increasing to 7.6% at sites with a backlog of two years or more, and the surveyers expect that trend to continue. The bigger the site, the more attracted it is to get some CASE in the joint too: OS sites have been quicker to adopt the technology than DOS sites. Even when comparing DOS sites to OS sites with identical application backlogs, the larger, more skilful OS sites are more open to the new technology than the DOS users. The study finds that 3.3% of DOS sites with backlogs in excess of 24 months use CASE tools whereas 10.7% of OS sites with backlogs of more than two years use the tools. The analysis and design tools most commonly used at the mainframe sites surveyed are: Excelerator, from Index Technology; Design Aid, from Nastec; Stradis/Draw from McDonnell Douglas, and CASE 2000, from Nastec. Excelerator leads the market with a 51.9% share, followed by Design Aid at 11.2% Stradis/Draw with 7.5% and CASE 2000 at 4.1%. Focus did not say how many sites were polled to make up its sample.