We report the results from a pilot search for radio recombination line (RRL) emission at millimeter wavelengths in a small sample of pre-planetary nebulae (pPNe) and young PNe (yPNe) with emerging central ionized regions. Observations of the H30α, H31α, H39α, H41α, H48β, H49β, H51β, and H55γ lines at ∼1 and ∼3 mm have been performed with the IRAM 30 m radio telescope. These lines are excellent probes of the dense inner ( 150 au) and heavily obscured regions of these objects, where the yet unknown agents for PN-shaping originate. We detected mm-RRLs in three objects: CRL 618, MWC 922, and M 2-9. For CRL 618, the only pPN with previous published detections of H41α, H35α, and H30α emission, we find significant changes in the line profiles indicating that current observations are probing regions of the ionized wind with larger expansion velocities and mass-loss rate than ∼29 yr ago. In the case of MWC 922, we observe a drastic transition from single-peaked profiles at 3 mm (H39α and H41α) to double-peaked profiles at 1 mm (H31α and H30α), which is consistent with maser amplification of the highest frequency lines; the observed line profiles are compatible with rotation and expansion of the ionized gas, probably arranged in a disk+wind system around a ∼5-10 M central mass. In M 2-9, the mm-RRL emission appears to be tracing a recent mass outburst by one of the stars of the central binary system. We present the results from non-LTE line and continuum radiative transfer models, which enables us to constrain the structure, kinematics, and physical conditions (electron temperature and density) of the ionized cores of our sample. We find temperatures T e ∼ 6000-17 000 K, mean densities n e ∼ 10 5 -10 8 cm −3 , radial density gradients n e ∝ r −αn with α n ∼ 2-3.5, and motions with velocities of ∼10-30 km s −1 in the ionized wind regions traced by these mm-wavelength observations. We deduce mass-loss rates ofṀ pAGB ≈ 10 −6 -10 −7 M yr −1 , which are significantly higher than the values adopted by stellar evolution models currently in use and would result in a transition from the asymptotic giant branch to the PN phase faster than hitherto assumed.