Leveraged ETFs have been known to stray of course by not exactly tracking the underlying index they’re based on. In “Are ETFs More Risky Than You Think?” this area is explored a little more in detail:
Unlike mutual funds, which are managed to try to get the best return, exchange-traded funds are meant to accurately reflect an entire index of stocks. Investors pick ETFs when they want to invest in a broader group of stocks, not on individual companies, and they generally count on ETFs to do as well — or as poorly — as the broader indexes on which they’re based.
But it turns out that a growing number of ETFs are failing to do just that. In 2009, tracking error — the difference between the performance of an ETF and its index — widened significantly, according to a recent Morgan Stanley report. U.S.-listed ETFs saw an average tracking error of 1.25% in 2009, more than double the 0.52% in 2008. The biggest offenders? Sector and industry funds, global funds, commodity and fixed-income ETFs.
So why does it matter if your ETF doesn’t do what it was created to do? Well, investors choose ETFs, instead of mutual funds or individual stocks, in order to diversify their holdings — and because they believe a broader index might perform better than a managed fund. From that perspective, tracking error can equal higher risk. In other words, tracking error can be cool when an ETF outperforms its index, and less cool when it underperforms the index.
…
The big question is, how do you know when your ETF has gone astray? Every ETF calculates what’s called the intraday indicative value of its underlying index, which essentially reflects its net asset value. Investors can compare an ETF’s market price to that indicative value to see how well the ETF is tracking its benchmark, and can monitor that matchup over time using an online quote system like Google Finance.
Simply enter the symbol of your ETF with “IV” and Google Finance returns with a chart of the indicative value over the past five days. You can then add the regular ETF symbol to this chart and see at a glance how close the ETF is tracking its index. Most ETF sponsors, particularly the larger ones, provide information about the indicative value going back to the ETF’s inception.
…
As Alan Segars, investment management officer of The Provident Bank’s Wealth Management Department, puts it: “Is there a certain threshold where investors should immediately sell their ETFs? Hardly. Rather, portfolio managers need to factor tracking error into a holistic investment process.” After all, some tracking error is inevitable, but you want to be sure that your ETF’s deviation matches your investment goals.
In some cases, the tracking error might be worth the benefits, Klein says. “You may just realize that the performance slippage to the market is a modest cost to pay for the benefits of low cost, diversified and liquid access to otherwise difficult to enter markets,” he says.
I must say that in all of my dealings with ETFs, I have not found this to be a major issue. I do recall, however, when back testing the SimpleHedgeStrategy a couple of years ago, that SDS not always performed as advertised by not accomplishing its goal of 200% inverse performance of the S&P; 500 during certain periods.
Using Google Finance, I followed the above suggestion and plotted SDS against its Intraday Value Index: 
As you can see, the tracking is fairly accurate and the fluctuations minor for the period shown. I would expect that to be even less with non-leveraged ETFs.
I agree with the article in that modest slippage is acceptable considering the otherwise very worthwhile benefits of using ETFs for structuring a portfolio.