We present in this paper an exhaustive compilation of all published data of extragalactic Cepheids. We have checked every light curve in order to characterize the different types of Cepheid and detect potential overtone pulsators, or to estimate the quality of the data. This compilation of about 3000 photometric measurements will constitute a very useful tool for astronomers involved for instance in the extragalactic distance scale.

After the first release of HIPPARCOS data, Feast & Catchpole gave a new value to the zero-point of the visual Cepheid Period-Luminosity relation based on trigonometric parallaxes. Because of the large uncertainties on these parallaxes, the way in which individual measurements are weighted bears a crucial importance, and the discrepancy they show leads to the conclusion that the choice of the best weighting system can be provided through a Monte-Carlo simulation. On the basis of such a simulation it is shown that: - A cut in \pi or in \sigma_{\pi} / \pi introduces a strong bias. - The zero-point is more stable when only the brightest Cepheids are used. - The Feast & Catchpole weighting gives the best zero-point and the lowest dispersion. After correction, the adopted visual Period-Luminosity relation is: <M_V> = -2.77 log P - 1.44 \pm 0.05. Moreover, we extend this study to the photometric I-band (Cousins) and obtain: <M_I> = -3.05 log P - 1.81 \pm 0.09.

We show that the Cepheid PL relation is affected by a Malmquist type bias, the so-called population incompleteness bias. Its calculated slope appears shallower than the true one because of the cutoff in apparent magnitude resulting from the instrumental limiting magnitude. Furthermore, the use of the PL relation, even with the correct slope, leads to underestimation of distances which is not negligible. We confirm this finding by studying simulated PL relations, and we show that this bias may be as large as 0.2 or 0.3 magnitudes on distance modulus. We also test the efficiency of a cutoff in $\log P$ and show that it is a good way to minimize this bias. However, a correction of this is difficult as long as the completeness of the sample is not perfectly well established.

The Lyon-Meudon Extragalactic database (LEDA) gives a free access to the main astrophysical parameters for more than 100,000 galaxies. The most common names are compiled allowing users to recover quickly any galaxy. All these measured astrophysical parameters are first reduced to a common system according to well defined reduction formulae leading to mean homogeneized parameters. Further, these parameters are also transformed into corrected parameters from widely accepted models. For instance, raw 21-cm line widths are transformed into mean standard widths after correction for instrumental effect and then into maximum velocity rotation properly corrected for inclination and non-circular velocity. This paper presents the reduction formulae for each parameter: coordinates, morphological type and luminosity class, diameter and axis ratio, apparent magnitude (UBV, IR, HI) and colors, maximum velocity rotation and central velocity dispersion, radial velocity, mean surface brightness, distance modulus and absolute magnitude, and group membership. For each of these parameters intermediate quantities are given: galactic extinction, inclination, K-correction etc.. All these parameters are available from direct connexion to LEDA (telnet lmc.univ-lyon1.fr, login: leda, no passwd OR http://www-obs.univ-lyon1.fr/leda ) and distributed on a standard CD-ROM (PGC-ROM 1996) by the Observatoire de Lyon via the CNRS (mail to petit@obs.univ-lyon1.fr).