Neutrino-induced pion production on nuclear targets is the major inelastic channel in all present-day neutrino-oscillation experiments. It has to be understood quantitatively in order to be able to reconstruct the neutrino-energy at experiments such as MiniBooNE or K2K and T2K. We report here results of cross section calculations for both this channel and for quasielastic scattering within the semiclassical GiBUU method. This methods contains scattering, both elastic and inelastic, absorption and side-feeding of channels all in a unitary, common theoretical framework and code. We find that charged current quasielastic scattering (CCQE) and $1 pi$ production are closely entangled in actual experiments, due to final state interactions of the scattered nucleons on one hand and of the $Delta$ resonances and pions, on the other hand. We discuss the uncertainties in the elementary pion production cross sections from ANL and BNL. We find the surprising result that the recent $1 pi$ production cross section data from MiniBooNE are well described by calculations without any FSI. For higher energies we study the validity of the Bloom-Gilman quark-hadron duality for both electron- and neutrino-induced reactions. While this duality holds quite well for nucleon targets, for nuclear targets the average resonance contributions to the structure function $F_2$ are always lower than the DIS values. This result indicates a significant impact of nuclear effects on observables, reducing the cross section and structure functions by at least 30-40% and changing the form of various distributions.