To support the unrelenting demand of high spectral efficiency and gigabit data rates driven by fast developing smart applications and internet of things, heterogeneous cellular networks (HCNs) with the multiple-antenna configuration have been presented as promising paradigms. In this paper, the downlink transmission performances of K-tier HCNs with multiple-antenna configurations are analyzed, where base stations (BSs) in each tier may differ in terms of the spatial density, transmit power, cell-bias factor, and the number of transmit antennas. Particularly, the service success probability is analytically developed with the stochastic geometry, which can evaluate the transmission reliability and congestion. The impacts of the cell association and user scheduling on the service success probability are derived with closed-form expressions, and the spatial multiplexing gains from the multiple-antenna configuration are exploited. The Monte Carlo simulation results demonstrate that the derived expressions for the service success probability are matched well, and indicate that the proper number of antennas should be chosen, which is strictly related to the densities of BSs and users in each tier.