This paper investigates the impact of news media sentiment on financial market returns and volatility in the long-term. We hypothesize that the way the media formulate and present news to the public produces different perceptions and, thus, incurs different investor behavior. To analyze such framing effects we distinguish between optimistic and pessimistic news frames. We construct a monthly media sentiment indicator by taking the ratio of the number of newspaper articles that contain predetermined negative words to the number of newspaper articles that contain predetermined positive words in the headline and/or the lead paragraph. Our results indicate that pessimistic news media sentiment is positively related to global market volatility and negatively related to global market returns 12 to 24 months in advance. We show that our media sentiment indicator reflects very well the financial market crises and pricing bubbles over the past 20 years.