You mean it's rotation on it's axis (ie length of the day)?
Mostly yes, it's slowing down slightly but at a constant rate - it doesn't vary much during the year.
(ok on very very small scale it varies with every storm, snowfall and earthquake)
If you mean the speed around the sun - then this changes summer-winter with the different distance from the sun.
In a sense, it is. It changes very gradually. However, the rotation period becomes shorter as we look back in our Earth's history. The planet used to rotate a lot faster, because of the gravitational pull of the moon. The moon used to be revolving closer to Earth. The closer it was, the faster the revolution of the moon, and hence, the faster the rotation of the Earth. The moon is moving away, causing us to slow down as it does so. So, as time goes on, our rotation will slow down.
No. The Earth's rotation rate is not constant. The rotation rate undergoes secular, periodic, and seemingly random changes.
The secular changes result from the Earth transferring angular momentum to the Moon. This is a very real, measurable effect. The length of one day (midnight to midnight) is increasing by about 2.3 milliseconds per century.
The periodic effects result from seasonal changes such as snow and ice building up in the Northern hemisphere during the winter and melting during spring/summer. This reduces the Earth's moment of inertia tensor during Northern hemisphere winter and increases it during summer. Since the Northern hemisphere is mostly land and the Southern mostly water, it is the seasons in the Northern hemisphere that drive these changes.
Even little things like earthquakes have a measurable effect on the length of day. Scientists don't know how to predict all these little changes exactly, but they try. After the fact, they can report the measured changes. The International Earth rotation and Reference systems Service (IERS) publishes forecasts of and measured values of DUT1 (google that term) on a regular basis.
This graph shows the difference between UT1 and UTC. The horizontal axis is in years. The vertical axis is DUT1=UT1-UTC (in seconds). If you spliced the ends of each descending section together you would get a graph of the difference between UT1 and Atomic Time (less a constant; TAI-UTC is presently 34 seconds). Subtract another 32.184 seconds and you get the additive inverse of Delta-T, the difference between Terrestrial Time and UT1. Ignoring those leap second discontinuities, the graph isn't a straight line.
The US Naval Observatory provides the timing services for the IERS. Website: http://maia.usno.navy.mil [Broken].