In the Oxford Record Linkage Study population in 1970, seven hundred ninety-three persons were hospitalized for or died as a result of pneumonia. Thirty-six percent who survived and 49% who died had been discharged from hospital within the previous five years. For the period 1963 through 1970, cohort analysis determined the probability of subsequent readmission and/or death caused by pneumonia within the next five years for patients discharged with any condition and with high-risk conditions only. From this analysis, it was estimated that pneumococcal immunization of relatively few discharged patients would prevent each subsequent readmission and death from pneumococcal pneumonia. These results suggest that, in addition to age and underlying medical condition, patterns of previous hospital care can be used to identify many persons at increased risk of developing pneumonia. If current patterns of previous hospital care are similar to those found in Oxfordshire, physicians should consider giving pneumococcal vaccine to patients who are discharged from hospitals.