The reliable prediction of the pressure drop of packed beds requires an accurate estimation of the packed-bed porosity as its error propagation multiplies the occurring errors by a factor of about four. The commonly used correlations to determine porosity depend only on the tube to particle diameter ratio. To improve accuracy, the influence of the filling method and of material properties, in particular surface roughness, density, and restitution coefficient are investigated for random packings of spheres in cylindrical confining walls. Furthermore, statistical methods are applied to identify interconnecting effects between these influencing parameters.