I like this particular plot and the ability to pass a function to the stat_func keyword to quickly plot up and visualize relationships between variables, but there's one thing. How do I 'turn off' or not plot the marginal distribution axes?

It looks nice but sometime I don't want this feature.

For example using this code:

import numpy as np
import seaborn as sns

x = np.arange(100) + np.random.randn(100)*20
y = np.arange(100) + np.random.randn(100)*20
sns.jointplot(x, y, kind='reg')

How can I remove the kde subplots on the top and right hand side of the main axes?

You could use JointGrid directly:

from scipy import stats

g = sns.JointGrid(x, y, ratio=100)
g.plot_joint(sns.regplot)
g.annotate(stats.pearsonr)
g.ax_marg_x.set_axis_off()
g.ax_marg_y.set_axis_off()