Average pulse profiles are generally separable into several components whereas individual components can be described by a Gaussian shape. One generally distinguishes between core and conal components, although it is yet unclear whether core and conal components originate from the same (Lyne & Manchester 1988) or a different emission mechanism (Rankin 1983). Investigating this problem, we studied the behaviour of individual components observed in about 200 profiles of 18 pulsars measured at three different frequencies, i.e. at 1.4 GHz, 4.75 GHz and 10.55 GHz. We unambiguously determined the number, location, width, flux density and spectral index of individual components using a special technique (Kramer et al. 1994). The applied method ensures that only intrinsic features of the pulse profiles are fitted by comparing off-pulse noise statistics and post-fit residuals as a goodness-of-fit estimator. As a first result, we note that all high S/N pulsar profiles can be accurately described by using only a small number (usually ≤ 5) of gaussian components, although we also find an example of a seven component pulse shape (B0740–28). Fitted components determined for different observations of the same pulsar were averaged, producing a grand mean profile which is completely noise-free and thus suitable for a reliable flux density measurement of individual components, for an accurate determination of the pulse width to calculate opening angles, ρ, of the emission cone, or for use as a template for obtaining pulse arrival times.