Scaling issue with multiple bar plots

Hi.

I have 2 data sets, both as a function of the some vector containing time stamps.
One of the datasets is an extract of the other where only values below below a certain threshold are kept. Rest are "null" values.
I am plotting both series in same bar plot but with different colors. This is done in order to highlight the bars which are below the threshold value.
I hope i am making sense so far...

Here's the script:

figure(2)
bar(DateTime5,e,'b','EdgeColor','b')
hold on
bar(DateTime5,e2,'r','EdgeColor','r')
datetick
xlim([DateTime5(1) DateTime5(end)])
legend('Error in percentage','Values where production <20%')
hold off

This script has worked for me so far, but suddenly now - with another dataset, the plot looks wierd:
http://www.mikaelhostrup.dk/images/cap1.png

whereas for other data sets it looks as it should:
http://www.mikaelhostrup.dk/images/cap1.png

Can anyone see what went wrong?