Recently, there has been considerable interest in the use of nanofluids for enhancing thermal performance. It has been shown that carbon nanotubes (CNTs) are capable of enhancing the thermal performance of conventional working liquids. Although much work has been devoted on the impact of CNT concentrations on the thermo-physical properties of nanofluids, the effects of preparation methods on the stability, thermal conductivity and viscosity of CNT suspensions are not well understood. This study is focused on providing experimental data on the effects of ultrasonication, temperature and surfactant on the thermo-physical properties of multi-walled carbon nanotube (MWCNT) nanofluids. Three types of surfactants were used in the experiments, namely, gum arabic (GA), sodium dodecylbenzene sulfonate (SDBS) and sodium dodecyl sulfate (SDS). The thermal conductivity and viscosity of the nanofluid suspensions were measured at various temperatures. The results showed that the use of GA in the nanofluid leads to superior thermal conductivity compared to the use of SDBS and SDS. With distilled water as the base liquid, the samples were prepared with 0.5 wt.% MWCNTs and 0.25% GA and sonicated at various times. The results showed that the sonication time influences the thermal conductivity, viscosity and dispersion of nanofluids. The thermal conductivity of nanofluids was typically enhanced with an increase in temperature and sonication time. In the present study, the maximum thermal conductivity enhancement was found to be 22.31% (the ratio of 1.22) at temperature of 45°C and sonication time of 40 min. The viscosity of nanofluids exhibited non-Newtonian shear-thinning behaviour. It was found that the viscosity of MWCNT nanofluids increases to a maximum value at a sonication time of 7 min and subsequently decreases with a further increase in sonication time. The presented data clearly indicated that the viscosity and thermal conductivity of nanofluids are influenced by the sonication time. Image analysis was carried out using TEM in order to observe the dispersion characteristics of all samples. The findings revealed that the CNT agglomerates breakup with increasing sonication time. At high sonication times, all agglomerates disappear and the CNTs are fragmented and their mean length decreases.