import matplotlib
from numpy.random import randn
import matplotlib.pyplot as plt
from matplotlib.ticker import FuncFormatter

def to_percent(y, position):
    # Ignore the passed in position. This has the effect of scaling the default
    # tick locations.
    s = str(100 * y)

    # The percent symbol needs escaping in latex
    if matplotlib.rcParams['text.usetex'] == True:
        return s + r'$\%$'
    else:
        return s + '%'

x = randn(5000)

# Make a normed histogram. It'll be multiplied by 100 later.
plt.hist(x, bins=50, normed=True)

# Create the formatter using the function to_percent. This multiplies all the
# default labels by 100, making them all percentages
formatter = FuncFormatter(to_percent)

# Set the formatter
plt.gca().yaxis.set_major_formatter(formatter)

plt.show()