What statistical test should I use???? - (Jan/22/2010 )

Dear all,
I'm extremely confused and I admit that I am extremely weak in biostatistics.

Alright, I'm trying to perform statistical test on my data, which is the percentages of bacterial successfully invaded into cells.

Initially, I was using t-test (as most of the publications used), but I was told that t-test is not suitable for percentages data. I did try non-parametric test - Mann-Whitney U test, but I found that it is extremely weak and most of my data are not statistically significant.

Should I perform data transformation on my data and apply t-test?

Thank you.

Chi-square-test or binomial test should do the job...