There's nothing like the prospect of a mega-acquisition to get a CEO's pulse racing. Why is this? There are endless war stories illustrating how this type of large-scale spending often ends in lost value and tattered reputations.

Of course, there are megadeals that do work, though they tend to be the exceptions. Too often, it's hard to figure out why some transactions succeeded while so many others didn't. It doesn't help that research reports tend to make sweeping assessments that often lump individual industries, companies and deals into one big, messy bucket.

In a new study, McKinsey & Co. senior expert Werner Rehm and partners Robert Uhlaner and Andy West tackle this M&A conundrum -- why so many deals seemingly fail and why CEOs keep doing them anyway. In their report, "Taking a Longer-Term Look at M&A Value Creation," the McKinsey authors analyze the excess shareholder returns for 639 of the world's biggest nonbanking companies, which completed more than 15,000 deals between 1999 and 2010.

They look, in particular, at performance by industry and by five strategies: programmatic (serial smaller deals that add cumulative significant market share); selective (ad hoc); tactical (many small deals, but not generating cumulatively significant market share); large M&A (more than 30% of the acquirer's market share); and organic growth.

It should come as no surprise that the bigger a company gets, the more it relies on M&A to generate growth. What's interesting, though, is that the odds of success often hinge -- apart from skill -- on the strategy of choice. CEOs, it turns out, tend to muck up megadeals far more frequently than they do smaller ones. Looking at individual industries, McKinsey's data suggests that high-tech companies tend to be the worst at pulling off megadeals. Those that did these transactions in the study underperformed -- delivering returns that fell 6.7% below the industry index. In fact, high-tech CEOs managed to deliver clear value only when they pursued tactical acquisitions, in those instances handing shareholders 1.2% above the industry average. As for all those thousands of companies that favor smaller deals and are thus divorced from the glitter of M&A league tables -- here's some good news. These so-called programmatic CEOs typically do a superior job delivering value over those who hoard the limelight doing headline-grabbing big M&A.

That should be encouraging for the majority of dealmakers, at least in the U.S., where they tend to dominate deal volume most years. According to Standard & Poor's Capital IQ, U.S. public companies did $785 billion in total M&A volume for 2011, with 48% of that stemming from undisclosed deals -- typically transactions that are fairly small.

Most industries other than manufacturing, industrials and insurance-related companies favor this type of M&A strategy (programmatic), with the most successful being telecoms and materials (which both outperformed the benchmark by 4.5% for the 10-year period). Much of this comes down to skill and focus, as most programmatic acquirers "prioritize one or two markets or product areas where they can build businesses with leadership positions," the Mc­Kinsey authors explain.

Still, why are big deals so hard to pull off? McKinsey suggests that those that do succeed tend to be in mature, slower-growing industries, where they can more readily squeeze out industry excess capacity, improve performance and spend more time doing complicated integration work. For those in fast-growing industries like high tech, CEOs can lose sight of the bigger picture as they slog through messy integrations, often missing out on critical product or upgrade cycles. Moreover, CEOs can be lured too often to do a big deal just when market valuations are running high.

It's doubtful, though, that many CEOs will be tempted to lose their heads this year. Most are hunkering down, focusing on strategies that can deliver measurable value. Laura Whitley, head of global commercial banking at Bank of America Merrill Lynch in New York, says: "I believe that M&A activity is down [because] many companies are opting for a more controllable path -- dividends or share repurchases to provide a return to shareholders." She expects some 20% of the bank's clients to do a deal this year, but the focus will likely be strategic. Who knows, as CEOs afflicted with mega-M&A urges wait for markets to improve, they could end up using their downtime for a bit of overdue reflection on how to ensure that they get their next deal right.