I assume you mean your gel is not running straight, the most likely problem is that the gel is somewhat inconsistent or have small air bobbles. Another option is that your lower gel was not level so your stacking gel/running gel boundary is not straight and therefore different part of the band enter the running gel at different times resulting in an angled band, try to cover your lower gel with a little bit of 0.1% SDS during polymerization this helps both with eliminating bobbles and with getting a straight gel.
If there are bands aligned at different angles and you are not having reproducible results it could mean a procedural mistake each time you repeat the process. Try to ensure reproducible steps in methodology following established protocol.
If so, I agree with Yoav. Are you pouring a 2-layered gel with a lower resolving and upper stacking? Or are you making a gradient gel with several layers?
If using a 2-layer gel: I cover my lower resolving gel with 70% ethanol to help polymerization and keep with gel even and remove any bubbles present at the top of the lower gel. It is important to remove all the 70% ethanol (or whatever you use to cover your lower gel) before pouring the upper stacking gel. Also, when you insert the comb into the stacking gel, make sure there are absolutely no bubbles present. When you eventually remove the comb, check to make sure your wells are even at the bottom.
If you can use a precast gel with your system, try it and see if the problem goes away. If it does, the issue is with your poured gels. If not, the problem could be an issue with your transfer. Make sure your gel and membrane are pressed together very securely, with no bubbles and no discernible buffer between them.
I agree a lot with Daniel from U of Cal., with a few additional thoughts. Depending what you mean by different angles, I have seen bands produced to have different shapes due to accidently poking into the stacking gel when loading the sample in the well. In addition to Daniel's comment on the transfer, make sure you are squaring off the gel to the membrane. Be aware if the membrane moves while pressing out any air bubble with any rolling device.
Also, you may want to consider using a stain such as Ponceau S on the membrane after the transfer to identify how the protein bands are before going through with the western. If they are not good bands, you may want to consider re-running the gel again, instead of wasting time and resources.
Lastly, if you are pouring your gels, Amresco (thorugh VWR Scientific) has a gel solution that one can pour that does not need or uses a stacking gel. It is called Next Gel. One solution, no need to make a stacking gel.