The study was conducted in Kasulu District Council to assess the determinants of the living arrangements of older persons in rural areas of Tanzania. A multistage sampling technique was applied to obtain 379 older persons to whom a questionnaire was administered for data collection. Data analysis was done using the likelihood chi-square ratio test and multinomial logistic regression using IBM SPSS statistics and STATA. It was found that 38% of older persons lived with a spouse and children while only 14% lived alone. Household size was a strong predictor of all types of living arrangements for older persons implying that the addition of one person to the household of an older person would increase the possibility of an older person living with a spouse and children relative to living alone. The marital status of the relative with whom the older person would prefer to live with was also a strong predictor. Also, it was found that older persons who had sources of income-generation were more likely to live with their relatives than those who had no sources of personal income. An older person with no education or low education was more likely to live with a relative compared to an educated one, and the effects were statistically significant for all categories of living arrangements. Furthermore, it was also found that a never-married older person had a reduced chance of living with a spouse. It is recommended that the current elderly and aging policy which puts more emphasis on family care for elderly members should be revised.