The first round of empirical studies on the new growth theory by and large focused on cross-country examinations. These studies were devoted to verify either the augmented Solow-Swan model (Mankiw, Romer, Weil, 1992) or the endogenous growth theory (see, for example, Sala-i-Martin, 1997). These studies were devoted to verify either the augmented Solow-Swan model (Mankiw, Romer, Weil, 1992) or the endogenous growth theory (see, for example, Sala-i-Martin, 1997). Recently criticism has been raised against cross-country econometric studies, resulting from the assumption of equal values of parameters in the growth models under scrutiny for all examined countries. As A. Greiner, W. Semmler and G. Gong (2005), Chapter I, state, such an assumption may be misleading. Firstly, crosscountry examinations, by lumping together countries at different stages of development, may miss the thresholds of development. Secondly, these studies rely on imprecise measures of the economic variables involved, and the results are by far nonrobust. Furthermore, different institutional conditions, social infrastructure and preference parameters will make the countries heterogeneous. Due to this recently in verifying growth models more and more often time series techniques are applied (see, for example, Jones, 1995; Lau, Shin, 1997; Lau, 1999; Greiner, Semmler, Gong, 2005; Ha, Howitt, 2006). In particular in Lau, Shin (1997), Lau (1999) and Ha, Howitt (2006) econometric implications of different exogenous, semi-endogenous and endogenous growth models are discussed in the context of cointegration analysis and cointegration techniques are applied to examine these models. In this paper we take a similar viewpoint and use non-linear cointegration techniques to verify the neoclassical Solow-Swan growth model (see Solow, 1956; Swan, 1956) and the endogenous growth model of Romer (see Romer, 1986) for 6 countries: the United States, Great Britain, Japan, Holland, France and Germany. The relaxation of the linearity assumption in cointegration analysis concerns here the dynamics of the adjustment process to the long-term equilibrium relationship. Namely, it is admitted that the adjustment may be either asymmetric - different for positive and negative deviations from the equilibrium - or, alternatively, disproportional - different for large deviations, in case of which the correction of the disequilibrium is stronger, and for small deviations, for which the correction is weaker or there is no correction at all. Simulation analyses in Pippenger, Goering (2000) and Bruzda (2006), (2007) point that standard cointegration tests lack their power in the presence of non-linear adjustments. On the other hand, one may expect that economic fluctuations connected with business cycles may cause the adjustment to be of a non-linear nature. This motivates the use of cointegration tests relaxing the linearity assumption of an adjustment process in verifying growth models. Further in the text in Section 2 the growth models of Solow and Romer are briefly presented with special emphasis on their econometric implications. In Section 3 the methodological approach used in the empirical examination is described, while in Section 4 empirical results are presented. (fragment of text)