We've all made plenty of slippery slope arguments in our day, and we've all pooh-poohed plenty. Do these arguments make sense, and, if so, when? This article tries to go behind the metaphor of the slippery slope to the mechanisms by which one step today may make the next step more likely tomorrow. "Slippery slopes," I argue, can operate through several distinct mechanisms, which need to be discussed separately. And these mechanisms, it turns out, relate to rational ignorance, heuristics, path-dependence, the expressive effect of law, and multi-peaked preferences - important subjects that have received extensive attention recently, but that have not so far been linked to the slippery slope question. I suggest that slippery slopes may indeed sometimes happen (though they aren't logically inevitable). The flip response that "if we can draw a line today, we'll be able to draw the line tomorrow" is correct only if decisionmakers have firm and single-peaked preferences, and unbounded rationality. In the real world, where these conditions don't always hold, one decision can indeed help grease the slope to another, in various ways. And this can happen not just with judicial decisions - where slippery slopes relate in complex ways to the system of precedent - but also with legislative ones, where precedent is not supposed to play a formal role. Understanding the full range of slippery slope mechanisms can help us evaluate the risk of slippage, craft better arguments related to this risk, and perhaps minimize this risk.