This is all about drag. The current answer by xpda is the correct idea, but I thought it might be worth going into a little more detail, to explain exactly why there is "no suction to speak of pulling back on it" (in fact, I wouldn't say there is "no suction"...rather, "less suction" would be more apt).
Drag results from a difference in pressure between the regions of air in front and behind the moving object in question. As air hits a single cyclist (let's say a female), it stagnates (slows down, resulting in high pressure) and moves around her. At real-world velocities and scales (where the Reynolds number is many orders of magnitudes greater than unity), the air is travelling too fast to remain attached over the cyclist's non-streamlined body. The air on each side is therefore not able to "meet-up" again at the trailing edge of the cyclist, as would be the case on a more streamlined object like an airfoil. Instead, the air on each side separates, forming shear layers (narrow streams of high-gradient velocity) downstream either side of the cyclist. Immediately after separation, these are separate from one another, but not far downstream they begin to interact (due to opposite-signed vorticity drawing them together). This results in a "wake" region behind the cyclist, where the flow is turbulent, eddies are plentiful, and the pressure is low.
When a second cyclist (let's say a male) is placed behind the first, the wake is affected. The presence of the new cyclist disrupts the interaction of the shear layers. A small amount of air will be drawn inward to form a mini-wake between the two cyclists, but the majority will continue to travel past the second cyclist (and may even re-attach to him before separating again), before interacting and forming a wake behind him. It is clear then, that the end result is mutually beneficial for both: the front cyclist has an unchanged high pressure region in front of her, but a "less negative" low pressure region behind her; the second cyclist has a far lower pressure region in front of him, but an (essentially) unchanged pressure region behind him. The difference in pressure between front and rear for each cyclist is therefore significantly lower for both. They both experience a lower drag than they would alone, and are more efficient together than the sum of their parts! Note the true picture has many more three-dimensional and transient wake effects in play (many are still yet to be fully understood and much research is going into this as we speak!), but the general idea holds.
However, the reduction in drag for the lead cyclist is likely to be less than that experienced by the rear cyclist. The lead cyclist still has unchanged stagnation pressure to deal with - which is a pressure coefficient of, say, around +0.5 (averaged over her front body). The pressure coefficient acting on the rear, however, rarely approaches negative pressures of that magnitude - it might be somewhere around -0.2 (these are all very rough estimates, but I feel the point is made better with numbers involved). With the cyclist behind her, that may rise to around -0.1. So the lead cyclist will be experiencing a pressure difference of around 0.6, as opposed to 0.7 normally. The second cyclist however, will have that stagnation pressure in front of him cut significantly from +0.5 to around, say, +0.1. Let's assume the pressure behind him is the same as if he were alone - around -0.2. The second cyclist is therefore now experiencing a pressure difference of only 0.3 - around half the pressure difference (aka half the drag!) of the lead cyclist! This rough calculation (I've simplified things significantly to get the point across) shows that it still pays to be the second cyclist!