So far, there have been introduced many studies addressing the effects of the substrate on silicon-based integrated RF passive components, but none of them has addressed to what extent the substrate has influence on the RF front-end circuits and phenomena, such as noise-figure and phase-noise. In particular, there are many different models of spiral inductors on silicon, and even more substrate models of the very same inductors, all of them claiming to be the most-right representatives of their on-chip counterparts. When subjected to a verification in a circuit environment, for example in LC oscillators, those models often predict a dramatically different performance of the oscillators, being different phase-noise, voltage-swing, loop-gain and tuning-range, for a given power consumption. Therefore, this aspect of model verification is presented in this paper, showing that the design trajectory of LC-oscillators depends to an unallowably large extent on the inductor models used in the analysis. Also, it is shown that the phase-noise of a quasi-tapped oscillator is, contrary to common belief, not a monotonic function of a substrate resistivity.