Worldwide, home deliveries persist especially in developing countries; SSA inclusive. In Tanzania, the trend was 53% in 1991-1992; 47% in 1996; and 44% in 1999; while 47% in 2004. In Songea, home delivery trend was 44% in 2005; 40% in 2006; and 42% in 2007 though antenatal care attendance was above 95%. This trend raises concern. The study was guided by two research questions: what determines home deliveries, and what impact do home deliveries have on maternal and infant mortality. The study was conducted in Songea Rural District. It adopted cross-sectional design and used structured questionnaires to collect primary data. Secondary data were obtained from reviewed related literatures. The study involved 200 respondents from five wards and 10 villages. The wards and villages were purposively selected while simple random method was used in choosing respondents. The study report is presented using descriptive statistics namely, frequencies and percentages. The results indicate that application of herbs for facilitating labour was leading factor by 98% of the respondents with home deliveries. Other causes were low income of people (87%), bad condition of roads (90%), long distance to health facilities (88%), inadequate delivery services at health facilities (89%); lack of women’s decision making power (76%), timing problem (75%), lack of transport to health facilities (91%), cost at health facilities (90%), and experience in previous deliveries (84%). Of the 15 maternal deaths found; 12 occurred among mothers delivering at home compared to three that delivered at health facilities. Moreover, home deliveries were associated with six infant deaths compared to three in health facilities. Reversing the trend of home deliveries requires improvement of infrastructures, empowering women to make decisions, sensitisation against social norms which promote home deliveries, retaining skilled personnel in rural facilities, and equipping health facilities with delivery materials regularly.