In the wake of the financial crisis it has become clear that there is a need for macroprudential oversight in addition to the existing microprudential banking supervision. One of the lessons from the crisis is that the network structure of the banking system has to be taken into account to assess systemic risk. There exists, however, no analysis on the influence of the network topology on contagion in financial networks. This paper therefore compares contagion in Barabási-Albert (scale-free) with Watts-Strogatz (small-world) and random networks. A network model of banks, a firm- and household-sector as well as a central bank is used. Banks optimize a portfolio of risky investments and risk-free excess reserves according to their risk and liquidity preferences. They form a network via interbank loans and face a stochastic supply of household deposits. Contagion effects from the default of a large bank are studied in different network topologies. The results indicate that contagion is more severe in random and scale-free networks than in small-world networks. This situation changes when the central bank is not active in which case small-world networks are less stable than scale-free and random networks. It is also shown that interbank liquidity above a certain threshold leads to endogenous instability, regardless of the network topology. The results further indicate that network heterogeneity does not contribute to financial instability.