Some interesting studies have been done on this topic.
The Kapooka study (‘A Randomised Trial of Pre-exercise Stretching for Prevention of Lower-Limb Injury’, Medicine and Science in Sports and Exercise, Vol. 32(2), pp. 271-277, 2000) (Kapooka is a hell hole in NSW where all Australian Army recruits do their basic training) had a very large sample size and showed that stretching does not reduce total injury levels. But, a later study (Effect of static stretching on prevention of injuries for military recruits, 2003, Amako) showed that stretching reduces the likelihood of muscle related injuries.
Note the distinction between ‘total’ and ‘muscular.’
Further detail below:
From http://www.sportsinjurybulletin.com/archive/stretching-exercises.html
But does stretching during a warm-up really cut an athlete’s chances of getting hurt? To find out, researchers from the Kapooka Health Centre, the University of Sydney, and Charles Sturt University in Australia recently examined the effects of pre-exercise stretching on lower-limb injury over 11 weeks of training in 1538 subjects ranging in age from 17 to 35(2). This study was carried out with a particularly apt study group: army recruits undergoing basic training. Although army recruits are not necessarily elite athletes, they do undertake a rigidly controlled and strenuous programme of exercise during basic training, and they also sustain a high frequency of lower-limb injury(3). Thus, if stretching is really beneficial as an injury-preventer, one would expect to see its effects in a large group of military-service signees.
What the programme involved
The 1538 recruits were randomly divided into stretching (735 individuals) and non-stretching groups (803). The Australian researchers decided to utilise a stretching programme comparable to the type of routine employed by many athletes and thus settled on 20 seconds of stretching for each of six key lower-limb muscles or muscle groups (the gastrocnemius, soleus, hamstrings, quadriceps muscles, hip adductors, and hip flexors) during warm-up. The chosen form of muscle unkinking was static stretching, in which a limb or portion of a limb is moved to close to the limit of its range of motion and then held in this stretched position, without continuous motion or overall body movement. Static stretching carried out in 20-second dosages has been shown to be effective at increasing joint range of motion(4) and at reducing muscular resistance to applied stretch(5).
The static stretches were interspersed with jogging and side-stepping activity during the warm-ups; naturally, individuals in the control group performed only jogging and side-stepping, without a hint of static stretching. During the 11-week period, 40 actual workouts were completed by the recruits, adding up to 50 total hours of hard physical effort. The training was divided into route marching (10 hours), running (10.5 hours), obstacle-course workouts (12.5 hours), circuit training (7.5 hours), swimming (four hours), and battle training (5.5 hours).
The results
Over the course of the 11 weeks (and 60,000 total hours of training), 333 lower-limb injuries were recorded, 175 in the control group and 158 in the stretching recruits, which represented an overall injury rate of 5.5 injuries per 1000 hours of training. The three most common injuries were patellofemoral pain (67 cases), tibial stress fractures (56), and ankle sprains (46). As it turned out, stretching during warm-up had no statistically significant effect on the risk of injury, either for soft-tissue problems or bony disorders. Height and weight of the military personnel were also non-factors when it came to predicting injury.
Although pre-exercise stretching was totally unimportant from an injury standpoint, other easy-to-determine factors actually did a decent job of prognosticating who would get hurt. For example, age was a good predictor of injury (the older the athlete, the higher the injury frequency), and even a non-training factor such as date of enlistment worked better than pre-workout stretching in terms of injury prediction (recruits who enlisted later in the year were more than twice as likely to get hurt, compared with those who enlisted in January, February, or March). In addition, 20-metre shuttle-run time was an outstanding predictor (the faster the time, the lower the risk of injury), a relationship which suggested that overall fitness, not the presence or absence of pre-workout stretching had the paramount influence on injury occurrence (a simple test like the 20-metre shuttle run is considered to be a reliable fitness assessment, since it can be used to accurately predict VO2max and running capacity)