In tumor hyperthermia, effectively planning in advance and thus controlling in situ the heating dosage within the target region are rather critical for the success of a therapy. Many studies have simulated the temperature distribution during hyperthermia. However, most of them are based on fixed and known heat source distributions, which are generally very complex to compute. Besides, there is little information concerned the numerical analysis of temperature during magnetic hyperthermia loading with magnetic nanoparticles (MNPs), which has its specific heat source distribution features. Particularly, the parameters for different human tissues varied very much, which will cause a serious impact on the heat source and temperature distribution. This paper is aimed at investigating the effects of nonuniform tissue properties to the temperature prediction in magnetic nanohyperthermia and other possible effect factors including external EM field, MNP properties, tumor size and depth, surface cooling conditions, etc. It was found that the spatial heat source generated in the nonuniform model appears smaller than that in the uniform model. This is mainly resulted from the energy reflection when transmitting from fat to tumor and muscle under the same condition, while the temperature is higher on account of overall contribution of different parameters including tissue thermal conductivity, blood perfusion, density, heat capacity, and metabolic heat production rate, which also affect the temperature distribution apart from the heat source. Controlling the properties of the external EM field, MNPs and cooling water can acquire different temperature distributions. Tumors with different depths and sizes need specific plannings, which require as accurate as possible temperature prediction. The nonuniform model can be further improved to be applied in magnetic nanohyperthermia treatment planning and thus help optimize the surgical procedures.