Growing opposition to vaccination and free-riding opportunities are undermining the success of im- munization programs giving rise to disease outbreaks, as recently happened in the case of measles in Europe. We analyze how individual vaccination decisions are affected by and affect social norms and disease prevalence, determining thus the dynamic evolution of the overall immunization coverage and disease persistence. We show that long run disease eradication or persistence depends on how key socio- health-economic parameters (i.e., the vaccine confidence level, the degree of conformism to social norms, and the degree of disease concern) relate to key epidemiological parameters (i.e., the basic reproduction number). Depending on the parameters configuration, alternative outcomes may arise, such as unique and multiple equilibria, monotonic and fluctuating trajectories, chaotic dynamics. We also show that the deterministic approximation of stochastic dynamics may lead to misleading conclusions about the true outcome because of the metastable properties of the stochastic system. Public policy by affecting key socio-health-economic parameters may play a fundamental role in ruling out some undesired outcomes and ensuring eradication. A calibration based on the recent measles outbreak in Italy confirms that our setup is able to well describe real world behavioral and epidemiological outcomes.