17 80 Low-dose contin¬ uous TMP-SMZ (Hughes et al'°)t 0 80 'Compared with no TMP-SMZ, x2 = .357, = 0.28. tCompared with high-dose TMP-SMZ, 2 = 4.15, = 0.02.Failure of patient compliance is unlikely because all oral medications were given with nursing supervision in the clinic or parent care facility.During our institutional outbreak of PCP, investigation by the Center for Disease Control noted an association between the length of hospitalization and development of pneumocystis pneumonia. In addition, a greater percentage of hematology staff, com¬ pared with other hospital staff, were seropositive for pneumocystis anti¬ bodies. Both observations suggest hos¬ pital transmission of the organism/ Recent animal studies confirm the spread of pneumocystis organisms by horizontal transmission."The results by Hughes et al·" employing a lower dose of TMP-SMZ (TMP 4 mg and SMZ 20 mg/kg/day) during the entire period of chemo¬ therapy are clearly superior to our results. Comparison of results be¬ tween the placebo and treated group in Hughes's series and between the no TMP-SMZ group and treated group in our series shows the superiority of long-term low-dose TMP-SMZ (Ta¬ ble). The attack rate of the groups who did not receive effective prophylaxis are quite similar: 17 of 80 (21%) of Hughes's placebo group compared with four of 17 (23%) of our nontreated group but, the time of onset of PCP differed. Our patients conform more to the general experi¬ ence of the investigators of the Chil¬ dren's Cancer Study Group \vho noted that the greatest frequency of pulmo¬ nary infections in 569 patients with ALL occurred within the first 100 days of treatment." The predictable period of high risk between 40 and 100 days from initiation of chemotherapy in our group was the basis for initiating the prophylaxis program at four weeks. Hughes's patients did not conform to this pattern since their infections occurred randomly throughout the entire period of treat¬ ment. This difference may be related to the intensity of therapy at differ¬ ent phases of the disease. Both Hughes' and our group have noted the correlation of increased PCP with the addition of more intense chemothera¬ py/'2 Since the incidence of PCP in our patients who received TMP-SMZ was not decreased, but simply delayed in onset for a period up to 170 days after initiation of chemotherapy and the rate of PCP in our non-ALL patients increased to 17%, we have abandoned the short-term high-dose TMP-SMZ as prophylaxis and are now evaluating the efficacy of low dose continuous