This study uses 2015/16 Tanzania Demographic and Health Survey (TDHS) data to estimate determinants of treatment seeking for childhood illness and the choice of health provider by employing logistic and multinomial probit model, respectively. Our empirical results from Binary logistic regression results show that treatment seeking for childhood illness is significantly related with mother’s occupation, household wealth status, distance to the health facilities, child’s age and place of residence. On the other hand, multinomial probit model results show that the choice of health provider is significantly related with mother’s occupation and access to mass media, household health insurance, household wealth status, and distance to the health facilities. Our results from both logistic and multinomial probit estimations are robust to alternative models’ specifications. In terms of policy implication, this study strongly recommends promotion of health insurance as well as creation of awareness on maternal and reproductive health to mothers. Moreover, the government should enhance, strengthen and ensure that health facilities are constructed close to households’ domicile and that these health facilities are provided with adequate services.