THE INMATE POPULATION PROVIDED A UNIQUE OPPORTUNITY FOR SUCH A STUDY IN THAT IT INCLUDED MEN WITH MINIMUM, MEDIUM, AND MAXIMUM SECURITY CLASSIFICATIONS, AND IS SO SITUATED THAT COMMUTING DISTANCES FOR VISITORS ARE GENERALLY SHORT. LETTERS AND VISITS FROM FAMILY MEMBERS, RELATIVES, AND FRIENDS WERE DETERMINED FOR A SAMPLE OF 120 INMATES FOR THE YEAR PRIOR TO THEIR PAROLE. RECIDIVISM WAS DEFINED AS RETURN TO PRISON WITHIN A 2 YEAR PERIOD FOLLOWING PAROLE. IT WAS HYPOTHESIZED THAT GREATER COMMUNITY CONTACT WOULD BE ASSOCIATED WITH LOWER RATES OF RECIDIVISM. OVER THE 1 YEAR PERIOD THE INMATE SAMPLE RECEIVED A TOTAL OF 4,225 LETTERS AND 3,975 VISITS. OF THE TOTAL SAMPLE 30 (24 PERCENT) WERE RECIDIVISTS AND 94 (76 PERCENT) WERE NONRECIDIVISTS. T-TESTS WERE USED TO COMPARE THE NONRECIDIVIST AND RECIDIVIST GROUPS ON THE FOLLOWING VARIABLES: NUMBER OF LETTERS RECEIVED FROM PARENTS, SIBLINGS, WIVES, CHILDREN, OTHER RELATIVES, FRIENDS, AND OVERALL LETTERS RECEIVED; AND NUMBER OF VISITS FROM EACH OF THESE PLUS OVERALL VISITS RECEIVED. NO SINGLE VARIABLE TESTED PROVED TO HAVE A SIGNIFICANT EFFECT ON RECIDIVISM; NOR DID COMBINING LETTERS AND VISITS INTO A SINGLE CATEGORY REVEAL SIGNIFICANT EFFECTS. HOWEVER DATA DID INDICATE A FAIRLY CONSISTENT TREND FAVORING THE HYPOTHESIZED RELATIONSHIP. SIGN TESTS USED IN A FURTHER ANALYSIS INDICATED A WEAK BUT SIGNIFICANT (AT THE .05 LEVEL) SUPPORT FOR THE HYPOTHESIS. METHODOLOGICAL PROBLEMS IN A STUDY OF THIS TYPE ARE DISCUSSED. STATISTICAL DATA ARE PRESENTED IN TABULAR FORM AND REFERENCES ARE INCLUDED. (JAP)