Sports scientists and trainers generally agree that the multidimensional training in tennis should start during early childhood in order to ultimately reach a professional playing standard. Evidence suggests that motor skills, including power, strength, agility, speed and explosive power, as well as mental strength and a highly developed neuromuscular coordinating ability are strongly correlated with the level of tournament performance. Turner&Dent (1996) found that 27% of all tennis injuries in junior players occur in the shoulder region. The shoulder girdle is prone to injury because of its ability to maximally accelerate and decelerate the arm while the arm maintains it maintains precise control over the racquet at ball contact. The purpose of this study was to determine whether the occurrence of shoulder injuries could be minimized in tennis players by following a specific exercise programme, focusing on the shoulder girdle. A total of 42 tennis players participated in this study. They were all aged between 14 and 18 years. Both males and females were used for the purpose of this study. All the players were training at the SA Tennis Performance Centre and the International Tennis Federation at the University of Pretoria. They were all elite tennis players practising daily and scheduled for standard major tournaments throughout the year. Each subject completed a questionnaire of his or her tennis and medical history. The players were then divided into a control group and an experimental group. Both groups completed a series of physical scientific tests, consisting of posture analysis, body composition, flexibility, functional strength of the upper body; and isokinetic power and endurance of the shoulder muscles. These tests were executed every 3 months over a 9-month period and the results of each battery of tests were used to adjust and upgrade the new programmes. The experimental group did specific preventative shoulder exercises 5 times a week in addition to their usual gymnasium programme twice a week, while the control group followed a normal strengthening programme twice a week. A medical doctor immediately evaluated any muscle stresses or pains throughout the year. At the end of the year the data was compared to determine the difference in injury occurrence between the two groups. There was a significant difference (p<0.05) in the distribution of the lean body mass with the Lean body mass at T1 being lower than the Lean body mass at T3 in the control group. In the experimental group the fat percentage showed a significant decrease (p<0.05) from T1 to T3. The distribution of the muscle percentage at T1 was significantly different (p<0.05) from the distribution of the muscle percentage at T3 in the experimental group with the muscle percentage at T1 being lower than the muscle percentage at T3. There was a significant difference between the control and experimental group for 1RM bench press (p<0.05) with the 1RM bench press measurements at T3 being lower for the control group than for the experimental group. Also, the 1RM bench press at T1 was lower than the 1RM bench press at T3 in the experimental group. The experimental group showed a significant increase from T1 to T3, peaking at T3 with the 1RM bench press. Results of the tests done to determine isokinetic muscle strength showed that a statistical significant correlation (p<0.05) was found with regard to the strength of the internal rotators of the non-dominant shoulder at T3, with the experimental group having a higher measurement than the control group. The internal rotators and external rotators of both the dominant and non-dominant shoulders were lower at T1 than at T3 in the experimental group (p<0.05). The external rotators of the non-dominant shoulder at T1 were lower than the external rotators of the non-dominant shoulder at T3 in the control group. Results of the tests done to determine flexibility showed a statistically significant difference with the internal rotators and external rotators of the dominant as well as the non-dominant shoulders being lower at T1 than at T3 in the experimental group. Also, the external rotators of the non-dominant shoulder of the control group were lower at T1 than at T3. Results of the tests done to determine posture showed that in the control group, 54.5% of the players had scoliosis at T1 as opposed to 40.9% at T3. In the experimental group 55% had scoliosis at T1 compared to the 30% at T3. In the experimental group, 55% of the players’ shoulder heights were not level at T1, compared to 30% at T3. 63.6% of the control group’s non-dominant shoulders were higher than the dominant shoulder at T1, compared to the 40.9% of subjects at T3. Among the subjects in the experimental group, 50% had a higher non-dominant shoulder and 5% a higher dominant shoulder at T1, compared to 25% and 5% respectively in the control group, at T3. Results of the tests done to determine the occurrence of injuries, showed that the subjects with no injuries in the control group stayed stable from T1 (54.5%) to T2 (54.5%) whereafter it increased to 59.1% at T3. The experimental group stayed stable from T1 (55.0%) to T2 (55.0%) where after it increased to 85% at T3. In the control group the percentage grade 1 and 2 injuries was 13.6% at T1, increasing to 18.2% at T2, and decreasing to 13.6% at T3. In the experimental group 15% of the subjects had grade 1 injuries at T1. This percentage increased to 30% at T2 where after it decreased to 15% at T3 again. The percentage of subjects with grade 2 injuries in the experimental group remained stable at 10.0% from T1 to T2. None of the subjects had grade 2 injuries at T3. In the control group 9% had grade 3 injuries at T1, with none at T2 and T3. In the experimental group the percentage of subjects with grade 3 injuries remained stable at 5.0% from T1 to T2. None of the subjects had grade 3 injuries at T3. In the control group 4.5% of subjects had grade 4 injuries at T1. This stayed more or less stable at T2 (4.6%) and increased to 9.1% at T3. In the experimental group 10.0% had grade 4 injuries at T1. None of the subjects had grade 4 injuries at either T2 or T3. In the control group 4.5% had grade 5 injuries at T1, none had it at T2, and 4.5% had it at T3. In the experimental group none of the subjects had grade 5 injuries at T1, T2 or T3. In the control group none of the subjects had grade 6 injuries at T1 or T3. At T2, however, 4.6% had grade 6 injuries. In the experimental group 5.0% of the subjects had grade 6 injuries at T1 and none had this type of injury at T2 or T3. In conclusion, the results indicate that a specifically designed exercise programme can help to diminish the risk of shoulder injuries in tennis players. It can also improve bi-lateral muscle strength in opposing muscle groups which are used in tennis.