Yes, despite randomization baseline characteristics can be different. This occurs especially with respect to categorical variables, when the categories are not balanced even after randomization. Mean age in the experimental and control group may not be significantly different. However, there may be significant difference in the distribution of population in each age group. Certain modifications like stratified randomization or CAR may help. But, they are not always helpful if there are multiple confounding factors. Such errors may be addressed by statistical techniques.

I agree with Kaushik. Yes, they can occur, because it is random procedure, instead of fixed assigning. You should adjust for it in your final comparison between the two strategies, especially if there is a difference in your primary outcome.