Unlike the Mercury, Gemini, Apollo and Space Shuttle programs, the future pace of NASA's programs for human and robotic space exploration will be driven by the space transportation system budget needed to lift heavy payloads to Earth escape velocity, with total heavy lift costs over 50 years of $100-$200 billion with present technology. This paper uses published cost and performance data to show that neither reusability nor high flight rate nor system improvements can be totally relied upon to lower launch costs significantly for heavy-lift boosters. Analysis of flight systems ranging from commercial aircraft to small expendable boosters to the Space Shuttle shows that launch cost correlates with propulsion system power, where launch costs ∼ (thrust power)1.09, a nearly-linear correlation that can be used to predict launch costs with reasonable accuracy. The case for horizontal launch rests on the large reduction in thrust power, and hence launch costs, possible with these systems. Subsonic, supersonic and launch with air liquefaction cycles are discussed. The paper quantifies important cost factors and predicts significantly lower costs for horizontally launched heavy-lift payloads than for vertical launch, with savings >$100 billion.