How to adjust the shape of a gamma distribution I have a dataset (vector) to which I would like to fit a gamma distribution (x), which given the shape seems reasonable. The charts below show the histograms of the two dataset: vector and x. 

The left is my sample set, the right is a random generated set. I did this as following; first: 
require(MASS)
fitdistr(vector,"gamma")
    shape     rate  
  4.84279   0.07493 
 (0.21693) (0.00354)

Then I generated the gamma distribution (right chart) as following:
hist(rgamma(10000, shape=4.84279, rate=0.07493),breaks=251)

I would like to change the shape in such a way that the peak gets slimmer, but I can not figure out how to do this. I have tried to modify the shape and rate, but neither of the two put me in the right direction. Does anybody have an advise on how to do this?
 A: The results don't necessarily show that a gamma distribution is a good fit. You might as well as assume that a normal distribution must fit well because you have estimated a mean and a standard deviation. 
Your graphical assessment is already showing that a good (if not necessarily the best) gamma shape for these data is systematically different from your data, and tweaking the parameters will at best tweak that, not solve the problem. 
I guess you need to explore other distributions. It is difficult to suggest precisely which, but my eyeballing from your graphs is that the gamma is not matching either the skewness or the kurtosis of your data. I would try lognormal. 
On graphical technique: I would rather look at a quantile-quantile plot than a histogram. Even with a histogram, simulating from the distribution is not as simple as just superimposing the density function, with parameter estimates plugged in, on the histogram. 
A footnote: What is going on near zero? You have a small spike there. Is there a substantive explanation? 
(LATER) Picking up a point made by @whuber: the left tail does seem to need more explanation. Perhaps you have a mixture here, a few values near zero, most values about above 25, and there's a story that perhaps should be matched by the analysis. Perhaps that gap is just sampling variation. If you know more, please tell us. 
