FORTUNE — The fears of another technology-stock bubble that prevailed in the first half of the year have faded away in the second half as a concerns about Europe’s financial stability took the speculative wind out of many Internet stocks. Few have gone public since June, and those that have ventured forth are trading below the highs they reached in their first days of trading.

So it’s strange to see one of the bad habits of the dot-com bubble resurface this year: research analysts, many of whom work for firms that have underwritten Internet stocks, are raising their ratings and price targets on stocks that — even at current levels — are wildly overvalued by any fundamental measure.

Take LinkedIn (LNKD). The professional social network went public in May and surged 138% to $122 on its first day of trading. A month later, however, LinkedIn had fallen back to $60 a share. That brought out a series of analyst reports from J.P. Morgan (JPM), Bank of America (BAC) and Morgan Stanley, LinkedIn’s three main underwriters. Each pounded the table, calling it “viral” and “transformative.”

At the time, I noted that justifying a stock with a triple-digit PE ratio required some stretching of logic, arcane financial metrics and impassioned rhetoric. But it worked, within a few days, Linked in was trading back at $102. And sure enough, after a few weeks, the stock began another descent. It finally hit a new low of $56 a share in late November, losing more than half its value in a little more than six months.

Analysts responded last week with a new wave of reports. On the same day, LinkedIn’s three main underwriters issued new reports that raised the firms’ stock ratings, price targets or earnings estimates. Each of them followed the standard format of weighing the positive against the negative. Each concluded that LinkedIn will soon be trading between $84 and $100. And each breezed right past the reality that those price targets defy any rational application of fundamental analysis.

But once again, it worked. By week’s end, LinkedIn’s shares were back near $72. At that valuation, LinkedIn is trading at 1,800 times the four cents a share it earned over the past 12 months. Even if you value LinkedIn against the 25 cents a share that the Street expects the company to earn in 2012, LinkedIn’s price-earnings ratio is 287.

LinkedIn’s underwriters aren’t alone. Analysts at other firms have recommended the stock in recent weeks. Barclays Capital (BCS) maintained a buy rating and $93 target, while Canaccord Genuity issued a buy rating and an $85 target. And unlike many of the hot-then-not IPOs of the dot-com years, LinkedIn has a solid business, a strong foothold in a growing market and the promise of years of growth. Revenue and membership have been doubling at an annual pace, international expansion is just starting and no formidable competitor has shown up yet in the professional networking space.

But the risk for investors buying LinkedIn at $70 today isn’t that the company is hurting, it’s that there is no rational explanation — beyond pure speculation — why the stock is so expensive. LinkedIn may in fact rise above $84 soon, but it’s a roll of the dice. It depends on further speculation and sheer volatility (exacerbated by the growing short interest on the stock).

And unlike the past several months, selling pressure is likely to grow heavy whenever LinkedIn approaches $100. That’s because of the tens of millions of new shares that have hit the market in the past month. On Nov. 17, LinkedIn had a secondary offering that issued 10 million new shares (the vast bulk of them sold by insiders). Two days later, the lock-up period ended for another 24 million shares, and in February, another 55 million shares will be unlocked.

Not all of the shares will be sold, of course, but the new shares could expand LinkedIn’s float from about 8% of outstanding shares at the time of its IPO to more than 20% today. That influx of shares is common, but it can also affect the supply and demand of LinkedIn’s shares.

Why would Wall Street issue reports justifying such an expensive stock only a decade after the great dot-com bust, when some analysts slapped misleadingly bullish recommendations on Internet IPOs? The answers aren’t necessarily nefarious.

Speculative rallies are the bane of securities analysts. No matter how sound their fundamental analysis, they look bad if they tell clients to sell a stock that keeps rising. For years, Netflix (NFLX) was a stock that kept on rising despite a perennially high valuation. But once it hit a speed bump this summer, the stock lost three-quarters of its value in a few months.

Underwriters, however, do have an incentive to promote recent tech IPOs. The sector has been one of the few that has reliably churned out offerings this year. Other IPOs like Zynga and Facebook entering the pipeline could see their prospects hurt by a slump in LinkedIn, the healthiest tech IPO of 2011.

Whatever the motivations for the bullish views on LinkedIn at its current price, there is something reckless about recommending a stock fueled largely by speculation, while downplaying the risks. Only one LinkedIn analyst seemed to be willing to address that risk: Morningstar’s Rick Summer, who admired LinkedIn’s operations but felt the stock was fairly valued at $47 a share, or 35% below its current level.

In a way, though, Summer is in good company. Only six months ago, LinkedIn’s underwriters valued the company at $45 a share when it set the offering price. The underwriting firms have yet to credibly explain why they believe the stock is now worth twice as much — and so soon.