In this study, a detailed investigation on the effects of entrance flow, buoyancy, inlet configurations, and fin geometries on heat transfer and friction factors in the entrance and fully developed regions of horizontal plain and micro-fin tubes under laminar, transition, and turbulent regimes is presented. The Reynolds numbers for the ethylene glycol-water mixtures throughout the experiments ranged from 800 to 25,000. The reliability of the experimental system was verified by plain tube heat transfer and friction factor experiments. Owing to the lack of information on entrance flow friction factors in the open literature, the effects of the inlet and heating on the entrance and fully developed flow friction factors are presented in this study. The results showed that the transition range was inlet-dependent and the effect of heating on friction factors was obvious in laminar and transitional regions. Furthermore, correlations predicting non-isothermal entrance and fully-developed friction factors in laminar and transitional regions for square-edged and re-entrant inlets were developed. From micro-fin tube heat transfer and friction factor test results, the transition from laminar to turbulent flow was clearly established and the transition Reynolds number was shown to be inlet- and fin geometry-dependent. The influence of buoyancy and entrance flow on heat transfer was obvious in the laminar region and the obvious effect of heating on the friction factor in the lower transitional region was observed. Correlations to predict forced and mixed convection heat transfer in laminar, transitional, and turbulent regions were developed. Also, correlations for predicting isothermal and non-isothermal entrance and fully-developed friction factors in laminar and transitional regions were developed. iv In this study, several soft computing methods were used to develop heat transfer correlations for plain and micro-fin tubes and determine optimal fin geometry for micro-fin tubes. The results showed that soft computing correlations are superior to correlations developed by the traditional least-squares method and that soft computing methods are applicable for the optimization problem of micro-fin tubes.