Probably quite a lot.. Charlie Cook points out:
"With three-quarters of the nation’s governorships and four-fifths of the state legislative seats up this year, almost all with four-year terms and last up in the pro-GOP 2010 and 2014 Obama midterm elections, Republicans now have huge exposure. In many states, like in the House, those district boundaries serve as protective walls against Republican losses. Tim Storey, the elections guru at the National Conference of State Legislatures, points out that from 1902 through 2014, the average midterm election outcome was a loss of 412 seats (including the previous odd-year’s elections); since World War II the losses have been a bit lower, 334 seats. Storey estimates that with a generic-congressional-ballot-test advantage of Democrats up by 6 points, that would likely translate into a gain of close to 500 state legislative seats nationwide for Democrats. Like in the U.S. House, the curve is asymmetric, the chances of over 500 are greater than under 400."
Democrats certainly have enough of a lead on the national generic to make a 500 seat gain for the party in state legislatures seem, if anything, like a pretty conservative estimate. Data indicate that there could be as many 1,000 Republican state legislative seats where Trump's approval rating is below 50 percent. That's a lot of targets.