Seven hundred and three cows were treated at drying off, using three main types of therapy: control, long-acting intramammaria and short-acting intramammaria. Selective dry cow therapy in infected quarters was used. Of the cows with one or two infected quarters one to six weeks before drying off, 57.7 per cent had one or more new infected quarters at drying off. Thus selective dry cow therapy on quarter basis determined from the results of samples taken one to six weeks before drying off would give "inadequate" therapy (i.e. new infection in non-treated quarters at drying off) for more than 50 per cent of the cows. For those cows given "adequate" therapy (no new infection at drying off) long-acting therapy, short-acting and control yielded 53.6%, 49.3% and 30.7% healthy cows respectively two to five weeks post partum. In cows with "inadequate" therapy the treatment groups yielded 42.9%, 52.9% and 29.7% healthy cows respectively. Our conclusion is that if long-acting intramammaria are to be used, all the quarters of infected cows should be treated, whereas if short-acting preparations are to be used, treatment can be restricted to infected quarters only.